# Nodespell Docs — full text > Official documentation for Nodespell, the node-based AI creative workspace for generating, compositing and finishing images, video and audio. This file contains every guide on https://docs.nodespell.com in navigation order. --- # What is Nodespell? > Nodespell is a browser-based node-graph workspace for generating and finishing images, video, audio and text with AI models, compositing and a Timeline. Source: https://docs.nodespell.com/start/what-is-nodespell/ · Last updated: 2026-11-10 Nodespell is a browser-based creative workspace where you generate and finish images, video, audio and text by connecting nodes in a graph. Every AI model, edit and effect is a node, every result stays attached to the node that made it, and you can change one step and rerun only what depends on it. You open Nodespell in a web browser at [app.nodespell.com](https://app.nodespell.com). There is nothing to install: AI model nodes run on Nodespell's servers, and your Projects, results and uploaded media are saved to your account. ## What can you make with Nodespell? Nodespell covers the whole path from a first idea to a finished file. The table lists what each kind of work looks like in the product. | You want to | In Nodespell you use | Start here | | ---------------------------------------------- | -------------------------------------------------------------------------------------------------------- | ------------------------------------------------------- | | Generate an image from a prompt | **Image Generator**, **Cinematic Shot** and other image Creative Tools | [Make your first image](https://docs.nodespell.com/start/first-image/) | | Edit, upscale, relight or cut out an image | **Image Editor**, **Image Upscaler**, **Relight**, **Variations**, **Remove Background** | [Creative Tools](https://docs.nodespell.com/creative-tools/) | | Animate a still or generate video from text | **Image to Video**, **Text to Video**, **Reference Images to Video**, **Extend Video**, **Modify Video** | [Image to video tutorial](https://docs.nodespell.com/start/first-image-to-video/) | | Generate voice, music and sound effects | **Voice Generator**, **Music Generator**, **Sound Effects Generator**, **Speak** | [Creative Tools](https://docs.nodespell.com/creative-tools/) | | Combine layers, key, color-correct and refocus | Compositor nodes such as Merge, Keyer, Color Correct, Curves, Defocus and ZDefocus | [Compositing](https://docs.nodespell.com/edit/compositing/) | | Cut clips and audio together | The **Timeline** editor in the **Design** panel | [Timeline](https://docs.nodespell.com/edit/timeline/) | | Ask for help or have changes made for you | The in-product **Agent**, shown in a panel labeled **Assistant** | [Agent](https://docs.nodespell.com/agent/) | The list of available models changes over time. The model picker and the Library panel in the app show what is available now, and [What's new](https://docs.nodespell.com/whats-new/) records each release. ## Who is Nodespell for? Nodespell is built for people who make images and moving pictures and want control over each step: artists, designers, motion and VFX artists, and small studios. Two parts of the product serve different starting points: - **Create** is a guided workspace. You pick a Creative Tool such as **Image Generator**, choose a model, write a prompt and select **Generate**. It suits someone who has never used a node editor. - **Graph** is the full node editor. You add, connect, organize and run nodes directly. It suits people who already think in node trees, such as compositors and technical artists. Both modes edit the same Project, so you can start in Create and open the result in Graph at any time. See [Choose a workspace mode](https://docs.nodespell.com/start/workspace-modes/). Teams can share Projects, Assets and, on the Studio plan, a shared credit pool. See [Teams](https://docs.nodespell.com/teams/). ## How is Nodespell different from a prompt-box generator? A prompt-box generator gives you a text field and a gallery of outputs. Nodespell keeps the steps that produced a result as visible, editable nodes. | In a prompt-box generator | In Nodespell | | -------------------------------------- | ---------------------------------------------------------------------------------------------------------------------- | | Each output is a separate download | Each output is a **take** stored on the node that made it, and you choose which take the rest of the graph uses | | Changing one step means starting again | You change one node, and nodes downstream reuse their existing takes until you choose to run them again | | One model per screen | Many models from several makers can sit in one graph, feeding each other | | Finishing happens in another app | Compositing, color correction, masks, depth effects, keyframes and a Timeline are nodes and panels in the same Project | See [Takes](https://docs.nodespell.com/graph/takes/) and [Run nodes](https://docs.nodespell.com/graph/run-nodes/) for how reuse works. ## How is Nodespell different from other node-based tools? Nodespell uses the same idea as other node editors (data flows from outputs to inputs along connections), with these differences: - **It runs in the browser.** AI model nodes run on Nodespell's servers, so you do not need a local GPU, model downloads or a Python setup. - **AI models and finishing tools share one graph.** A generated image can feed a compositor, a Color Correct node and then a video model without leaving the Project. - **Guided and direct editing work on the same nodes.** A Creative Tool is a guided presentation of a model node, not a separate copy. See [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/). - **Costs are visible before you run.** Paid plans show a credit estimate for the selected model and settings before you start a run. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## What does Nodespell not do? - Nodespell does not run AI models on your own computer. Model nodes need an internet connection and use credits. - A free account can open Projects and work with your own media, but video generation is part of paid plans. - Nodespell is not a general-purpose code or automation environment. Its nodes are built for images, video, audio and text. ## Frequently asked questions ### Do I need to install anything to use Nodespell? No. Nodespell runs in a web browser at app.nodespell.com. AI model nodes run on Nodespell's servers, and your work is saved to your account. ### Do I have to learn node graphs before I can make something? No. The **Create** workspace lets you choose a Creative Tool, a model and a prompt and select **Generate** without touching the graph. The graph is built for you alongside, and you can open it later in **Graph** mode. ### Does generating with AI models cost money? Running AI model nodes uses credits. Paid plans include monthly credits, and a free account's credits do not renew. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ### Can several people work in the same Project? Teams can share Projects and Assets in a Team workspace. See [Team projects](https://docs.nodespell.com/teams/team-projects/) for how shared Projects work. ## Related - [Make your first image](https://docs.nodespell.com/start/first-image/): Generate an image in Create, step by step. - [Core concepts](https://docs.nodespell.com/start/core-concepts/): Projects, nodes, connections, runs, takes and credits in one page. - [Choose a workspace mode](https://docs.nodespell.com/start/workspace-modes/): When to use Create, Graph or Immersive. - [Tour the interface](https://docs.nodespell.com/start/interface/): Top bar, panels, docks, layouts and the command palette. --- # Make your first image in Create > Generate your first image in Nodespell Create with Image Generator, check the credit estimate, inspect the result, and continue from it. Source: https://docs.nodespell.com/start/first-image/ · Last updated: 2026-11-10 This tutorial makes a first image in Nodespell with **Image Generator** in the **Create** workspace. You open a Project, choose a model, write a prompt, select **Generate**, and then inspect the result, which Nodespell keeps on an Image Generator node in the Project's graph so you can continue from it. You need a Nodespell account with credits. Generating uses credits; the [plans and credits guide](https://docs.nodespell.com/account/plans-and-credits/) explains balances. The whole tutorial takes about five minutes plus generation time. ## Before you start: open a Project in Create Create works inside a Project. A Project is the saved workspace that holds your graph, results and layout. 1. Sign in at [app.nodespell.com](https://app.nodespell.com). The dashboard lists your Projects. 2. Select **New Project**. If your dashboard is empty, select the **Make an image** or **Start a blank graph** card instead; each card creates a new Project. 3. Look at the left end of the top bar. The **Workspace mode** selector shows the current mode name: **Create**, **Graph** or **Immersive**. 4. If it does not say **Create**, open the selector and choose **Create** from the **Workspace** menu. New accounts start in Create. The mode you choose is saved as your preference and applies to every Project you open. > **Note:** Create needs a window at least 1080 pixels wide. In a narrower window Nodespell shows **Create needs a wider workspace** with an **Open Graph** button. Widen the window to return to Create. ## What is on the Create screen? The Create screen has three areas from left to right. | Area | What it contains | | ----------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Create tools rail | **All tools**, then pinned tools: **Image Generator**, **Image to Video**, **Audio Generator**, **Image Upscaler**. Below a divider: **Assistant** (the [Agent](https://docs.nodespell.com/agent/)) and **Assets**. | | Composer | The selected tool's name, its model card, its controls, the **Generations** stepper and the main action button, such as **Generate**. | | Graph | The Project's real node graph. Every generation you start in Create adds or updates a node here. | Select the expand button at the bottom of the rail to show tool names next to the icons. The rail can expand only when the window is wide enough. ## Generate an image 1. **Choose Image Generator.** Select **Image Generator** in the Create tools rail. The composer heading changes to **Image Generator** with the description "Create a new image from a prompt." 2. **Check the model.** The model card at the top of the composer shows the selected model. Keep it for your first run. To change it, select the card to open **Choose model**, or use the **Previous model** and **Next model** arrows beside it. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). 3. **Write the prompt.** Type a description in the **Prompt** field. Name the subject, the setting, the lighting and the framing, for example: "A foggy harbor at dawn, fishing boats at anchor, soft golden light, 35mm film still". 4. **Review the other controls.** Controls such as **Aspect ratio**, **Resolution** and **Output format** appear only when the selected model supports them. Less common controls are inside **More settings**. Keep the defaults for now. 5. **Keep Generations at 1.** The **Generations** stepper next to the action button sets how many images Nodespell makes each time you select **Generate**, from 1 to 4. 6. **Check the cost.** On paid plans, the **Generate** button shows a credit estimate with a coin icon. On a free account the estimate slot shows a locked **Cost estimate** that explains paid plans show it. 7. **Select Generate.** Nodespell checks the required fields, submits the run, adds an **Image Generator** node to the graph, and selects it. If a required field is empty or out of range, the field shows a message such as "This value is required." Fix it and select **Generate** again. > **First-run guide:** On a new account, a **Make your first image** card can appear in the composer. **Try this example** fills in a sample prompt and, when one is available, a quick, low-cost model. **Another example** swaps the prompt. You can reopen the card from **Help** > **Show first-run guide**. ## Watch the run and inspect the result While the node runs, the **Results** section at the top of the composer shows its state: **Submitting**, **Running**, then **Complete** or **Failed**. When you generate more than one image, a **Generation batch** list shows each generation's progress. When the image is ready: | Control | What it does | | --------------------------------------------- | --------------------------------------------------------------------------------------------------------- | | Result preview | Shows the selected result inside the composer. | | **View fullscreen** | Opens the result in the fullscreen viewer. | | **Show older result** / **Show newer result** | Moves through earlier and later results on the same node. A counter shows the position, such as "2 of 3". | | **Continue with this result** | Starts the next tool from this exact image. See the next section. | The image also appears on the **Image Generator** node in the graph on the right. Each result is a [take](https://docs.nodespell.com/graph/takes/): selecting **Generate** again with the same node selected adds a new take instead of replacing the old one. To start a separate image instead of another take on the same node, click an empty area of the graph to clear the selection. The composer then holds a new draft, and the next **Generate** adds a new node. ## Continue from your result Select **Continue with this result** under the preview. The **Continue with this result** dialog offers: - **Image Upscaler**: "Create a larger, refined version of this image." - **Image to Video**: "Use this image as the starting frame for a video." Nodespell keeps the exact image you chose by adding a visible Asset node to the graph, named **Selected historical result**, and connects it to the new tool. New takes on the original node do not change it. The next tutorial, [Make your first image-to-video clip](https://docs.nodespell.com/start/first-image-to-video/), uses this path. ## What if something goes wrong? | Message | What it means and what to do | | -------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------- | | **Not enough credits** | The run needs more credits than your balance. The sheet shows the estimate and your balance. Top up or choose a plan. | | A model name followed by **is part of paid plans** | The selected model needs a paid plan. Your prompt and settings stay as they are. Choose another model or a plan. | | **The Create action could not be completed.** | The run was not submitted. Try again; if it keeps failing, see [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/). | | **Failed** in Results | The run was submitted but the model returned an error. Check the prompt and settings, then generate again. | ## Frequently asked questions ### Where does Nodespell save my generated image? Nodespell saves each result on the node that produced it, inside your Project. You can browse earlier results with **Show older result** and choose which take the graph uses. To download or reuse the image elsewhere, see [Export](https://docs.nodespell.com/edit/export/) and the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ### How many images can I generate at once in Create? The **Generations** stepper allows 1 to 4 generations each time you select **Generate**. With more than one generation, the estimate on the button covers the whole batch. ### Why do the controls change when I pick another model? Create shows only the controls the selected model supports. Switching models can add, remove or rename controls such as **Aspect ratio** or **Resolution**, and Nodespell tells you which settings could not be carried over. ### Can I see the graph Create built for me? Yes. The graph on the right of the Create screen is the Project's real graph. Open it full size by choosing **Graph** in the **Workspace mode** selector, or select **Open unrestricted Graph editing** in the composer header. ## Related - [Make your first image-to-video clip](https://docs.nodespell.com/start/first-image-to-video/) — animate the image you just made. - [Build your first graph](https://docs.nodespell.com/start/first-graph/) — build the same kind of work directly in Graph. - [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) — every control and model option. - [Takes](https://docs.nodespell.com/graph/takes/) — browse and choose between results. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) — how estimates and balances work. --- # Build your first connected graph > Build a two-node Nodespell graph in Graph mode, connect Image Generator to Image Upscaler, run it, and see how takes are reused on the next run. Source: https://docs.nodespell.com/start/first-graph/ · Last updated: 2026-11-10 This tutorial builds a small connected graph in Nodespell's **Graph** workspace: an **Image Generator** node feeding an **Image Upscaler** node. You add nodes with the **Tab menu**, connect them, set a prompt in **Parameters**, run the graph, and then change one step to see which nodes run again and which reuse their saved takes. You need a Project and credits for two model runs. If you have not made an image yet, [Make your first image](https://docs.nodespell.com/start/first-image/) is the shorter introduction. ## Switch the Project to Graph 1. Open a Project from the dashboard, or select **New Project**. 2. Open the **Workspace mode** selector at the left end of the top bar and choose **Graph**. 3. Find the **Graph** panel, the large canvas with a dotted or lined grid. If it is not visible, select **+** on any pane's tab bar and choose **Graph** under **Essential**. 4. Make sure a **Parameters** panel is open as well. If not, select **+** and choose **Parameters**. Graph mode remembers your panel layout. The [interface tour](https://docs.nodespell.com/start/interface/) explains panes, tabs and docks. ## How do you move around the canvas? Use whichever input you have. Every action has a trackpad or keyboard path; none needs a middle mouse button. | Action | Trackpad | Mouse | Keyboard | | -------------------- | --------------- | ------------------ | ----------- | | Pan | Two-finger drag | Middle-button drag | Arrow keys | | Zoom | Pinch | Scroll wheel | `=` and `-` | | Frame selected nodes | — | — | `F` | | Frame all nodes | — | — | `A` | The status bar's **Hints** area shows the pan and zoom gestures for the input device Nodespell detects. More navigation is in the [Graph Editor guide](https://docs.nodespell.com/graph/graph-editor/). ## Add Image Generator with the Tab menu 1. Move the pointer over an empty area of the Graph panel. 2. Press `Tab`. The **Tab menu** opens at the pointer with a search field. 3. Type `Image Generator`. 4. Select **Image Generator** in the results, or press `Enter`. The node appears where the menu opened. If you prefer the pointer, right-click (two-finger click on a trackpad) an empty area of the canvas and choose **Search all nodes**, which opens the same search. The same menu lists **Creative tools** you can place with one click. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/) for the **Node Browser** and other ways to add nodes. **Image Generator** is a [Creative Tool](https://docs.nodespell.com/creative-tools/): a guided presentation of a model node. It starts with a default model, which you can change later. ## Write the prompt in Parameters 1. Click the **Image Generator** node to select it. 2. In the **Parameters** panel, find the **Prompt** field and type a description, for example "A ceramic teapot on a linen tablecloth, soft window light, studio product photo". 3. Leave the other controls at their defaults. The **Parameters** panel always shows the controls of the selected node. Changes save as you type. See [Parameters](https://docs.nodespell.com/graph/parameters/). ## Add Image Upscaler and connect it 1. Keep **Image Generator** selected. 2. Press `Tab`, type `Image Upscaler`, and press `Enter`. 3. Nodespell places **Image Upscaler** next to the selected node and connects the image output of **Image Generator** to the image input of **Image Upscaler**. A connection runs from an output socket on one node to an input socket on another. To connect two nodes yourself, drag from the output socket to a compatible input socket. To remove a connection, double-click it. Nodespell rejects a connection when the output and input types do not match; see [Connections](https://docs.nodespell.com/graph/connections/). ## Run the graph 1. Click **Image Upscaler** to select it. 2. Look at the floating run controls near the top of the Graph panel. With a node selected, the primary button shows what a run will do, for example **Generate 2**: one take for Image Upscaler and a first take for Image Generator upstream, which has none yet. 3. Hover the button to see which nodes will generate, reuse or update. 4. Select the button, or press `Cmd/Ctrl+Shift+R` (**Run Selected Tasks**). Both nodes show progress while they run. When they finish, each node displays its output. Select a result and press `Space` to open it in the fullscreen viewer; press `Esc` to close it. > **Caution:** A run that generates takes uses credits. On paid plans the run controls show the estimate before you start. ## Change one step and run again This is the part that makes a graph different from a single prompt. 1. Select **Image Generator** and change a word in the **Prompt**. 2. Select **Image Generator** only, and run it again. The button reads **Generate**. Image Generator makes a new take. 3. Select **Image Upscaler**. It still uses the image it upscaled before, because downstream nodes reuse their existing takes until you run them. 4. Run **Image Upscaler** to upscale the new image. Each node keeps every take it has made. With one node selected, the **Takes** button beside the run controls lets you preview earlier takes and choose the one the graph uses. See [Takes](https://docs.nodespell.com/graph/takes/) and [Run nodes](https://docs.nodespell.com/graph/run-nodes/). ## Frequently asked questions ### Why did running Image Upscaler also generate Image Generator? A node cannot run without its inputs. When an upstream model node has no take yet, the run plan generates that first take too. The hover preview and the button label, such as **Generate 2**, show this before you start. ### Does changing a prompt rerun everything downstream? No. Nodespell does not rerun nodes automatically. Downstream nodes keep using their existing takes until you run them, so nothing is charged until you choose to. ### How do I delete a node or a connection? Select a node and press `Backspace` or `Delete`. To remove a connection, double-click it, or select a node and press `X` to disconnect all its connections. ### Can I see this graph in Create? Yes. Switch the **Workspace mode** selector to **Create** and select a Creative Tool node in the graph. The composer then shows that node's model and controls. ## Related - [Graph Editor](https://docs.nodespell.com/graph/graph-editor/) — canvas navigation, selection and the minimap. - [Add nodes](https://docs.nodespell.com/graph/add-nodes/) — the Tab menu, Quick Add and the Node Browser. - [Connections](https://docs.nodespell.com/graph/connections/) — sockets, types and multiple inputs. - [Run nodes](https://docs.nodespell.com/graph/run-nodes/) — run plans, reuse and costs. - [Takes](https://docs.nodespell.com/graph/takes/) — compare and choose results. --- # Make your first image-to-video clip > Turn a still image into a video clip in Nodespell with the Image to Video Creative Tool, from a generated result or an uploaded image. Source: https://docs.nodespell.com/start/first-image-to-video/ · Last updated: 2026-11-10 This tutorial turns a still image into a short video clip with Nodespell's **Image to Video** Creative Tool in the **Create** workspace. You start from an image you generated or uploaded, describe the motion in a prompt, check the duration and cost, select **Generate**, and review the clip, which Nodespell keeps on an Image to Video node in your graph. > **Paid plans:** Video generation is part of Nodespell's paid plans. On a free account, selecting **Generate** opens a sheet titled **Video generation is part of paid plans**, and your prompt and settings stay as they are. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Choose your starting image Image to Video needs one image to animate. Pick the path that matches what you have. **From a generated image** Use this path after [Make your first image](https://docs.nodespell.com/start/first-image/). 1. In **Create**, select the **Image Generator** node that holds your image, or select **Image Generator** in the rail and generate one. 2. In the **Results** section of the composer, use **Show older result** and **Show newer result** until the image you want is showing. 3. Select **Continue with this result**. 4. In the **Continue with this result** dialog, choose **Image to Video**. The composer switches to **Image to Video** with your image already in its image field. A notice explains that the tool "will use the exact selected result through a visible Asset node": Nodespell adds an Asset node named **Selected historical result** to the graph, so later takes on the Image Generator node do not change your video's source. Select **Remove** in that notice if you change your mind. **From your own image** 1. In **Create**, select **Image to Video** in the Create tools rail. 2. Find the image field. Most models label it **Start frame**; some label it **Source image**. 3. Select **Choose assets** in that field. A dialog opens on your Nodespell Assets. 4. Select an image you uploaded before, or upload a new file from the dialog. 5. Select **Use selection**. Supported upload formats and sizes are listed in [Supported media](https://docs.nodespell.com/reference/supported-media/). You can also upload from the **Assets** tool in the Create rail; see [Upload and organize assets](https://docs.nodespell.com/assets/upload-and-organize/). ## Choose a model The model card at the top of the composer shows the selected video model. Keep the default for a first clip, or select the card to open **Choose model** and compare models. Each model supports different durations, resolutions and inputs, and the composer changes to match. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). If you switch to a model that cannot accept a setting you already entered, Nodespell tells you which settings were adjusted or could not be carried over. ## Describe the motion Type into the **Prompt** field what should move and how the camera behaves. The image already supplies the subject and look, so spend the words on motion: - The subject's action: "the boats rock gently, gulls cross the frame". - The camera: "slow push-in", "static camera", "slow pan left". - Changes over time: "the fog thins as the sun rises". Avoid describing a different scene from the image. The model treats the image as the first frame. ## Set duration and other controls The controls below the prompt depend on the model. The table lists the common ones and the section they appear in. | Control | Section | What it sets | | ------------------ | ---------- | ---------------------------------------------------------------------------------------- | | **Start frame** | References | The image the clip starts from. | | **End frame** | References | An optional image the clip should end on, for models that support first and last frames. | | **Duration** | Motion | Clip length. The credit estimate updates when you change it. | | **Resolution** | Output | Output size, when the model offers a choice. | | **Aspect ratio** | Output | Frame shape, when the model offers a choice. | | **Generate audio** | Audio | Whether the model also generates sound, for models that can. | Less common controls, such as **Seed**, appear under **Settings** or inside the collapsible **More settings** group. When you change controls inside **More settings**, its header shows how many you changed. ## Check the estimate and generate 1. Keep **Generations** at `1`. Each extra generation is a separate video run. 2. On paid plans, read the credit estimate shown with a coin icon on the **Generate** button. It updates as you change the model, duration or resolution. 3. Select **Generate**. Nodespell checks required fields first. If the image field is empty, it is marked as required and nothing is submitted. If your balance is too low, a **Not enough credits** sheet shows how many credits the run needs and how many you have. ## Review the clip Video runs take longer than image runs. The **Results** section shows **Submitting**, **Running** and then **Complete** or **Failed**. The node in the graph shows the same progress. When the clip is ready: - Play it in the result preview in the composer. - Select **View fullscreen** for a larger player. - Use **Show older result** and **Show newer result** to compare clips if you generate again. To try a different motion, edit the **Prompt** and select **Generate** again with the same node selected. Each clip is kept as a separate [take](https://docs.nodespell.com/graph/takes/) on the node. ## Where to go from the clip - Open the graph in **Graph** mode to see the chain: Image Generator, the **Selected historical result** Asset node, and Image to Video. - Lengthen the shot with **Extend Video**, or raise its resolution with **Video Upscaler**. See [Creative Tools](https://docs.nodespell.com/creative-tools/). - Add music or sound and cut it with other clips in the [Timeline](https://docs.nodespell.com/edit/timeline/). - Download the file; see [Export](https://docs.nodespell.com/edit/export/). ## Frequently asked questions ### Can I control the last frame of the clip? Some Image to Video models accept an **End frame** as well as a **Start frame**. The field appears only when the selected model supports it. ### Why does my clip change the subject or style of my image? Video models interpret the prompt as well as the image. Describe motion and camera rather than a new scene, and keep the prompt consistent with what the image shows. Trying another model can also help. ### Does generating a new clip replace the old one? No. Every clip is saved as a take on the Image to Video node. Browse them with **Show older result** and **Show newer result**, and choose which one the graph uses in [Takes](https://docs.nodespell.com/graph/takes/). ### Why is Image to Video asking for a paid plan? Video generation is part of paid plans. A free account can still open the tool, set it up and connect it in the graph, but it cannot run video models. ## Related - [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) — every control and model option. - [Make your first image](https://docs.nodespell.com/start/first-image/) — create the source image. - [Takes](https://docs.nodespell.com/graph/takes/) — compare clips and choose one. - [Timeline](https://docs.nodespell.com/edit/timeline/) — assemble clips and audio. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) — estimates and video access. --- # Choose a workspace mode: Create, Graph or Immersive > Compare Nodespell's three workspace modes, Create, Graph and Immersive, and learn how to switch between them and what each one shows. Source: https://docs.nodespell.com/start/workspace-modes/ · Last updated: 2026-11-10 Nodespell has three workspace modes that show the same Project in different ways. **Create** is a guided screen for generating with Creative Tools, **Graph** is the full node editor with panels you arrange yourself, and **Immersive** gives the graph the whole window with floating tool rails. Switching modes never changes your graph or results. ## How do you switch workspace mode? 1. Open a Project. 2. Select the **Workspace mode** selector at the left end of the top bar. It shows the current mode's name. 3. Choose **Create**, **Graph** or **Immersive** from the **Workspace** menu. In Immersive the top bar is collapsed. Select the **Main menu** button in the top-left corner to expand it, then use the same selector. Your mode is saved as a personal preference, so the next Project you open uses the same mode. New accounts start in Create. If Nodespell cannot save the change, it shows **Workspace mode could not be saved** and keeps the previous mode. ## Compare the three modes | | **Create** | **Graph** | **Immersive** | | ------------------------ | -------------------------------------------------- | ------------------------------------------ | -------------------------------------------------------------------------------------- | | Best for | Generating with a guided tool, one model at a time | Building, connecting and finishing graphs | Working on the graph with the most canvas space | | Main layout | Tools rail, composer, and the graph on the right | Panes, tabs and edge docks you arrange | One full-window graph with left and right tool rails | | Add nodes | Choose a tool, or use the graph on the right | `Tab` menu, right-click menu, Node Browser | **Add node**, **Creative tools**, **Models**, **Library** on the left rail, plus `Tab` | | Edit settings | Composer controls for the selected tool or node | **Parameters** panel | **Parameters** quick panel on the right rail when a node is selected | | Save and restore layouts | No | Yes, from the **View** menu | No | | Edge docks | No | Yes | No | | Floating panel windows | No | Yes | Yes, but they cannot be docked | | Status bar | Shown | Shown | Transparent overlay without the console | | Minimum window width | 1080 pixels | None stated | None stated | ## When should you use Create? Choose **Create** when you know the task, such as "make an image" or "animate this still", and want the controls for it in one place. The Create screen has: - A **Create tools** rail on the left with **All tools**, the pinned tools **Image Generator**, **Image to Video**, **Audio Generator** and **Image Upscaler**, and the helpers **Assistant** (the [Agent](https://docs.nodespell.com/agent/)) and **Assets**. - A composer with the tool's model card, its controls grouped into sections such as **Prompt**, **References**, **Output**, **Motion**, **Audio** and **Settings**, a **More settings** group, the **Generations** stepper and the action button, such as **Generate**, **Upscale** or **Extend**. - The Project's real graph on the right, where every run adds or updates a node. When you select a compatible model node in the graph, the composer edits that node. When you select several nodes, the composer shows **Choose one node**, because Create edits one model node at a time. A node whose model is not presented in Create shows **This node uses Graph controls** with an **Open in Graph** button. Create is not available for every Project. The **Create** item in the menu is disabled with **Unavailable for this project** when you are viewing a read-only Project or a saved snapshot, among other cases. Choose **Graph** to work in those Projects. Start with [Make your first image](https://docs.nodespell.com/start/first-image/). ## When should you use Graph? Choose **Graph** for direct node work: adding any node, wiring several models together, compositing, keyframing and organizing large graphs. Graph is the only mode with the full panel system: - Panes you can split, resize and maximize, each holding tabs. - Edge docks on the four sides of the window, toggled with `Cmd/Ctrl+1` to `Cmd/Ctrl+4`. - Saved layouts and **Built-in Layouts** in the **View** menu. - The **Version Bar**, shown or hidden from the **View** menu. See [Build your first graph](https://docs.nodespell.com/start/first-graph/) and the [interface tour](https://docs.nodespell.com/start/interface/). ## When should you use Immersive? Choose **Immersive** when you want the graph to fill the window. Immersive shows one graph editor behind a collapsed top bar and a transparent status bar, with two floating rails: | Rail | Items | | ----- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Left | **Creative tools**, **Add node** (the full Tab menu), **Prompt**, **Add asset**, **Group**, **Library**, and **Models** with image, video, audio and realtime effect views | | Right | **Parameters** and **Design** while a node is selected, then **Assets**, **Assistant** and **Console** | Hover a rail to see labels. Right-rail items open as quick panels beside the rail; each panel has buttons to detach it into a floating window, maximize it or close it, and `Esc` closes it. The **Console** item shows a badge when there are errors or warnings you have not looked at. Immersive keeps your Graph layout untouched. When you switch back to Graph, your panes and docks are where you left them. ## Frequently asked questions ### Does switching workspace mode change my Project? No. All three modes show the same graph, nodes and takes. Only the layout around the graph changes, and your Graph layout is preserved while you use Create or Immersive. ### Why is Create greyed out in the Workspace menu? Create is disabled with **Unavailable for this project** when the open Project cannot be edited from Create, for example a read-only Project or a saved snapshot. Use **Graph** instead. ### Why does Create say it needs a wider workspace? Create needs a window at least 1080 pixels wide. In a narrower window it shows **Create needs a wider workspace** and an **Open Graph** button. Widen the window or continue in Graph. ### Is the workspace mode saved per Project? No. The mode is saved as your personal preference and applies to every Project you open. ## Related - [Make your first image](https://docs.nodespell.com/start/first-image/) — a first run in Create. - [Build your first graph](https://docs.nodespell.com/start/first-graph/) — a first graph in Graph mode. - [Tour the interface](https://docs.nodespell.com/start/interface/) — panes, docks, layouts and menus. - [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/) — the same tools as graph nodes. - [Panels reference](https://docs.nodespell.com/reference/panels/) — every panel and its purpose. --- # Tour the Nodespell workspace interface > Find your way around the Nodespell workspace, including the top bar and menus, panels and tabs, docks and layouts, the status bar and the command palette. Source: https://docs.nodespell.com/start/interface/ · Last updated: 2026-11-10 The Nodespell workspace is the screen you see inside an open Project. It has a top bar with menus and status, a central area of panels such as **Graph**, **Parameters**, **Assets** and **Design**, and a status bar along the bottom. In **Graph** mode you can split, dock, float, color and save these panels as layouts. Most of this page describes **Graph** mode, which has the full panel system. **Create** and **Immersive** use fixed layouts; see [Choose a workspace mode](https://docs.nodespell.com/start/workspace-modes/). ## What is in the top bar? The top bar runs across the top of the window, from left to right: | Element | What it does | | ---------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Workspace mode** selector | Shows the current mode and switches between **Create**, **Graph** and **Immersive**. | | Menus | **File**, **Edit**, **View**, **Layout**, **Window**, **Help** and **Account**. The **Window** menu is hidden in Create. | | Project name and save status | Shows the Project name and whether your latest changes are **Saved**, **Saving…**, **Retrying**, **Not saved** or **Needs action**. Select it for details. | | Command palette button | Opens the command palette. The tooltip shows the shortcut `Cmd/Ctrl+K`. | | **Messages** | Opens messages from Nodespell, with a count of unread ones. | | Credits | Shows your remaining credits. Select it for the balance, renewal date and which credits this Project uses (personal or a team's shared pool). | Save status and recovery are covered in [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/). Credits are covered in [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## What is in each menu? | Menu | Main items | | ----------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | **File** | **New Project**, **Open ...**, **Recent Projects**, **Save As**, **Save Version ...**, **Versions**, **Go To Live Version**, **Close Project** | | **Edit** | **Undo** (`Cmd/Ctrl+Z`), **Redo** (`Cmd/Ctrl+Shift+Z`), **Copy**, **Paste**, **Preferences** (`Cmd/Ctrl+,`) | | **View** | In Graph: **Save Current Layout**, **Edit Layout**, **Layouts**, **Built-in Layouts**, **Show Version Bar**. In all modes: **Show Status Bar** (not in Immersive), **Show Command Palette**, **Show Quick Node Creation Menu** (`Tab`). In Graph: **Toggle Top Dock**, **Toggle Bottom Dock**, **Toggle Left Dock**, **Toggle Right Dock** | | **Layout** | **Tidy up** and **Flow direction** for the current Level and Project defaults. See [Organize a graph](https://docs.nodespell.com/graph/organize/). | | **Window** | **New Panel Window**, **Open Windows**, **Collect Offscreen Windows**, **Minimize All Windows**, **Restore All Windows** | | **Help** | Documentation, What's new, tutorials, support and feedback. See [Help menu and What's new](#help-menu-and-whats-new). | | **Account** | **My account**, **Assets**, **Cookie settings**, **Log out** | **Save Version ...** creates a Project Snapshot. See [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/). ## Panels, panes and tabs A **panel** is one tool, such as **Graph** or **Parameters**. A **pane** is a rectangle of the workspace that holds one or more panels as **tabs**. Only one tab in a pane is visible at a time; select a tab to show it. Add a panel to a pane: 1. Select **+** at the end of the pane's tab bar. 2. Choose a panel from the menu. It is grouped into **Essential** (**Graph**, **Parameters**, **Assets**, **Design**), **Library** (**Library**, **My Nodes**), **Automation** (**Tasks**, **Assistant**) and **Advanced** (**References**, **Tree View**, **Console**). Every panel is described in the [panels reference](https://docs.nodespell.com/reference/panels/). Right-click a tab (two-finger click on a trackpad) for its menu: | Tab menu item | What it does | | ------------------------------------ | ---------------------------------------------------------------------------------------- | | **Maximize pane** / **Restore pane** | Gives the pane the whole workspace, or returns it to its place. | | **Tear off** | Moves the panel into its own floating window. | | **Send to edge dock** | Moves the panel into the **Top dock**, **Bottom dock**, **Left dock** or **Right dock**. | | **Duplicate** | Opens a second copy of the panel. | | **Rename** | Changes the tab's title. | | **Colour** | Tints the tab. See [Color your panel tabs](#color-your-panel-tabs). | | **Move** | **Move to first**, **Move Left**, **Move Right**, **Move to last**. | | **Help** | Opens the documentation for that panel. | | **Close** | Closes the panel. | Double-click a tab, or an empty part of the tab bar, to maximize its pane; double-click again to restore it. The pane's tab bar menu also has **Auto-hide tab bar**, which hides the tab bar until you point at it. ## Split, move and resize panes Drag a tab to rearrange the workspace. While you drag, a preview shows where the panel will land and a label names the result. | Drop the tab on | Result | | --------------------------------------------- | -------------------------------------------------------------------------------------------- | | Another tab bar | **Place tab here**: the panel joins that pane at that position. | | The middle of another pane | **Add to this pane**: the panel joins that pane as the last tab. | | The left, right, top or bottom edge of a pane | **Split left**, **Split right**, **Split top** or **Split bottom**: the pane divides in two. | | A dock strip at a window edge | **Dock to left edge** (or top, bottom, right). | | Anywhere while holding `Shift` | **Float as a window**. | Drag the divider between two panes to resize them. Select the **Delete Pane** button in a pane's corner to remove the pane. ## Floating windows A floating window is a panel outside the pane layout, which you can move anywhere over the workspace. - Create one from **Window** > **New Panel Window**, with **Tear off** on a tab, or by holding `Shift` while you drag a tab. - Each window has **Minimize**, **Maximize** and **Close** controls in its title bar, and **Return to workspace** to dock it again. You can also drag the title bar over a tab bar or a pane edge. - **Window** > **Open Windows** lists every window, including ones moved off screen, with **Move Into View**. **Collect Offscreen Windows** brings them all back. - The **Windows** button at the left of the status bar offers the same list. In Immersive, floating windows work the same way but cannot be docked. ## Edge docks An edge dock is a panel area that slides in from one edge of the window: top, bottom, left or right. Edge docks are available in Graph mode. | Action | How | | --------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------- | | Show or hide a dock | **View** > **Toggle Top Dock** (and Bottom, Left, Right), or `Cmd/Ctrl+1`, `Cmd/Ctrl+2`, `Cmd/Ctrl+3`, `Cmd/Ctrl+4` for top, bottom, left and right | | Put a panel in a dock | Right-click its tab and choose **Send to edge dock**, or drag the tab onto a dock strip | | Add a new panel to a dock | Select **+** in the dock | | Change a dock's size | **Dock size**: **Compact (25%)**, **Half (50%)**, **Large (75%)** or **Full (95%)** | | Return a panel to the panes | **Return to workspace** | An empty dock says **No panels in this dock.** ## Color your panel tabs Give a tab its own color so you can tell similar panels apart, such as two **Graph** panels showing different Levels. 1. Right-click the tab. 2. Choose **Colour**. 3. Choose one of 16 swatches (from **Red** to **Pink**), **Custom colour** for any color, or **No colour** to return to the default. Tab colors are saved with the layout. ## Save and restore layouts A layout is the arrangement of panes, tabs, docks and tab colors. Layouts are available in Graph mode. | Action | Menu path | | ---------------------------- | ---------------------------------------------------------------------- | | Save the current arrangement | **View** > **Save Current Layout**, enter a **Name**, select **Save** | | Switch to a saved layout | **View** > **Layouts** > the layout name | | Use a built-in layout | **View** > **Built-in Layouts** > **AI - Essential** or **Power User** | | Rename or delete layouts | **View** > **Edit Layout** | The command palette also lists each layout as **Restore Layout:** followed by its name. ## The status bar The status bar runs along the bottom of the workspace. Hide or show it with **View** > **Hide Status Bar** or **Show Status Bar**. | Area | What it shows | | --------------- | -------------------------------------------------------------------------------------------------------------------------------------------- | | **Windows** | A menu of floating windows. Not shown in Create. | | **Hints** | Shortcuts for what you are doing now, such as the pan and zoom gestures for your detected input device (**Mouse**, **Touchpad** and others). | | Console summary | The latest log message and counts of new errors and warnings. Select it to open a **Console** window. Not shown in Immersive. | ## Use the command palette The command palette is a searchable list of commands, settings, nodes and tools. 1. Press `Cmd/Ctrl+K`, select the command button in the top bar, or choose **View** > **Show Command Palette**. 2. Type part of a command name. The field reads "Type a command or search...". 3. Select a result, or use the arrow keys and press `Enter`. Commands are grouped, for example **File**, **Panels**, **Node Actions**, **Selection**, **Navigation & View**, **Tasks**, **User Preferences** and **Help**. Many show their keyboard shortcut. Examples: **Run Selected Tasks**, **Tidy up**, **Toggle Minimap**, **Toggle Theme**, **Open New Panel** and **Add Parameters Panel**. Panel commands are not offered in Create. All shortcuts are listed in [Shortcuts](https://docs.nodespell.com/reference/shortcuts/). ## Help menu and What's new The **Help** menu collects documentation, updates and support: | Item | What it does | | --------------------------------- | -------------------------------------------------------------------------------------------------------- | | **Help** | Opens this documentation site. | | **What's new** | Opens the release notes. A **New** label and a dot on **Help** mean there are updates you have not seen. | | **Getting started tutorials** | Opens **Learn the essentials**, nine short lessons on core actions. | | **Show first-run guide** | Reopens the first-image guide in Create. Shown when Create is available. | | **Show suggestions** | Turns contextual tips on or off. | | **Email** | Copies the support email address. | | **Report a problem…** | Opens a message to support with technical details filled in. | | **Discord** / **YouTube channel** | Opens the Nodespell community and video channel. | | **Submit feedback** | Sends a **Feature Request**, **Bug Report**, **General Feedback** or **Question**. | | **About** | Opens **Help and about**, with links, legal documents and **Copy details** for support requests. | When a release ships, a **What's new** card can appear with **See what's new** and **Dismiss**. You can always find it again under **Help** > **What's new**, or under **More** > **What's new** on the dashboard. Every release is also listed on [What's new](https://docs.nodespell.com/whats-new/). For contacting support, see [Support](https://docs.nodespell.com/account/support/). ## Frequently asked questions ### How do I get a closed panel back? Select **+** on any tab bar and choose the panel again, or press `Cmd/Ctrl+K` and search for **Add** followed by the panel name, such as **Add Parameters Panel**. ### A floating window disappeared off the screen. How do I get it back? Choose **Window** > **Collect Offscreen Windows**, or open **Window** > **Open Windows**, find the window marked **(out)**, and choose **Move Into View**. ### Why can't I find layouts or docks? Layouts and edge docks are part of Graph mode. In Create and Immersive the **View** menu does not show them. Switch the **Workspace mode** selector to **Graph**. ### Are layouts saved per Project? Saved layouts belong to your account and appear under **View** > **Layouts** in any Project. Use **View** > **Edit Layout** to rename or delete them. ## Related - [Choose a workspace mode](https://docs.nodespell.com/start/workspace-modes/) — Create, Graph and Immersive. - [Panels reference](https://docs.nodespell.com/reference/panels/) — what every panel does. - [Shortcuts](https://docs.nodespell.com/reference/shortcuts/) — all keyboard shortcuts. - [Account settings](https://docs.nodespell.com/account/settings/) — Preferences, theme and language. - [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/) — save status and Needs attention. --- # Nodespell core concepts in one page > Learn the Nodespell mental model in one page, covering Projects, nodes, connections, parameters, runs, takes, Levels, Assets, credits and Creative Tools. Source: https://docs.nodespell.com/start/core-concepts/ · Last updated: 2026-11-10 Nodespell is built on a small set of ideas. A Project holds a graph of nodes; nodes pass media and data to each other through connections; running a node produces a take, which is kept; and AI model runs use credits. Each idea is defined once below, with a link to its full guide. ## How the pieces fit together | Concept | One-line definition | Full guide | | ---------------- | ------------------------------------------------------------------- | ----------------------------------------------------------- | | Project | The saved workspace that holds a graph, its takes and its layout | [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/) | | Node | One step: an AI model, an edit, an effect, an input or an organizer | [Add nodes](https://docs.nodespell.com/graph/add-nodes/) | | Connection | A link from one node's output to another node's input | [Connections](https://docs.nodespell.com/graph/connections/) | | Parameter | A setting on a node, such as a prompt, a duration or a blur size | [Parameters](https://docs.nodespell.com/graph/parameters/) | | Run | Executing selected nodes and whatever they need upstream | [Run nodes](https://docs.nodespell.com/graph/run-nodes/) | | Take | One saved result of a node's run | [Takes](https://docs.nodespell.com/graph/takes/) | | Level and Folder | A graph inside a node, for structure and reuse | [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/) | | Asset | Uploaded or generated media stored in Nodespell | [Asset Manager](https://docs.nodespell.com/assets/asset-manager/) | | Credits | The usage balance that AI model runs draw from | [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) | | Creative Tool | A guided presentation of an AI model node for a familiar task | [Creative Tools](https://docs.nodespell.com/creative-tools/) | ## Project A Project is the unit you open, save and share in Nodespell. It holds the node graph, every take its nodes have made, and the panel layout. Nodespell saves changes automatically; the top bar shows **Saved**, **Saving…** or a problem status. **File** > **Save Version ...** keeps a named Project Snapshot you can return to. Projects live in your personal workspace or a Team workspace. ## Node A node is one step in the graph. Nodes come in several kinds: - **AI model nodes** call a specific model, such as an image or video generator. They run on Nodespell's servers and use credits. - **Processing nodes** edit or combine media, such as Merge, Color Correct or Transform. - **Input nodes** bring material in: an **Asset** node holds an image, video or audio file, and a **Prompt** node holds reusable text. - **Organizing nodes** structure the graph: **Group**, **Folder**, Level and Dot nodes. Every node has a name, input sockets on one side, output sockets on the other, and parameters. The Library panel in the app lists every node; see [Add nodes](https://docs.nodespell.com/graph/add-nodes/). ## Connection A connection carries data from an output socket to an input socket. Data always flows one way, from output to input. Nodespell accepts a connection only when the output's type fits the input, so an image output cannot feed an audio input. Some inputs accept several connections, such as a list of reference images. See [Connections](https://docs.nodespell.com/graph/connections/). ## Parameter A parameter is a value on a node that you set in the **Parameters** panel or, in Create, in the composer. Parameters include text (a prompt), numbers (a duration), choices (an aspect ratio), colors and media. Many parameters can also take an input connection or an [expression](https://docs.nodespell.com/graph/expressions/) instead of a typed value. ## Run A run executes the selected nodes. Before it starts, Nodespell builds a plan: which nodes generate new takes, which reuse takes they already have, and which only update. Hover the run button to see the plan; on paid plans it also shows a credit estimate. Nodes never rerun automatically when something upstream changes. You decide when to run, so nothing is charged until you do. Press `Cmd/Ctrl+Shift+R` for **Run Selected Tasks**, or select the run button above the graph. ## Take A take is one saved result of a node. Each run of a model node adds a take; earlier takes are kept, not overwritten. The node's output uses one take at a time: by default the newest successful one (**Follow latest**), or one you choose with **Use this take** and keep with **Keep this take**. Downstream nodes use whatever take their input node is using. See [Takes](https://docs.nodespell.com/graph/takes/). ## Level and Folder A Level is a node that contains its own graph. Press `Enter` on a Level to go inside and `U` to go back up. A **Folder** is a Level used only for organizing; it cannot run directly. Use **Convert to Folder** and **Convert to Level** to switch between them. See [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/). ## Asset An Asset is a media file stored in Nodespell: something you uploaded, or a result saved from a node. The **Assets** panel (the Asset Manager) lets you browse, search, tag, preview and organize Assets, and drag them into the graph as Asset nodes. Assets belong to your personal workspace (**Personal Assets**) or to a Team (**Team Assets**). See [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ## Credits Credits are the balance that AI model runs draw from. Different models use different amounts depending on the model, the settings and the output length. Paid plans include monthly credits; credits on a free account do not renew. Each take in a node's take list shows whether its credits were **charged**, **refunded**, **not charged** or are **charge pending**. Runs that only reuse existing takes are not charged. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Creative Tool and model node A **Creative Tool**, such as **Image Generator** or **Image to Video**, is a guided way to use AI model nodes for a familiar task. It shows a curated set of controls and lets you switch between suitable models. A **model node** is the same model shown with its own name and its full list of parameters. A Creative Tool changes the editing experience, not the model, its price, its results or its connections. On the canvas you can switch a node between **Show as Creative Tool** and **Show as** followed by the model name. Changing the model on a node that already has takes creates a new **branch** node so the original and its takes stay intact. See [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/) and [Branch and switch presentation](https://docs.nodespell.com/graph/branch-and-switch-presentation/). ## Frequently asked questions ### What is the difference between a take and an Asset? A take is a result stored on the node that made it, part of that node's history. An Asset is a media file in the Asset Manager that you can reuse anywhere. You can turn a chosen take into an Asset node in the graph, for example with **Continue with this result** in Create. ### Do I pay again when I reopen a Project? No. Opening a Project, browsing takes and running with existing takes does not use credits. Credits are used when a run generates new takes with an AI model. ### Is a Folder the same as a Level? No. Both contain a graph, but a Level can run as a node while a Folder is only for organizing and cannot run directly. ### What is the difference between a Workflow and a Project? A Project is the full workspace you open and edit. A Workflow is a reusable piece of graph saved with **Save As Workflow** that you can place into any Project. See [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/). ## Related - [What is Nodespell?](https://docs.nodespell.com/start/what-is-nodespell/) — what you can make and who it is for. - [Build your first graph](https://docs.nodespell.com/start/first-graph/) — see these concepts in a real graph. - [Glossary](https://docs.nodespell.com/reference/glossary/) — every Nodespell term defined. - [Run nodes](https://docs.nodespell.com/graph/run-nodes/) — run plans, reuse and costs. - [Takes](https://docs.nodespell.com/graph/takes/) — choose which result the graph uses. --- # Creative Tools in Nodespell > Learn what Nodespell Creative Tools are, compare all 19 tools by input and result, and see how each tool relates to the model nodes in your graph. Source: https://docs.nodespell.com/creative-tools/ · Last updated: 2026-11-10 Creative Tools are Nodespell's guided controls for familiar creative jobs, such as generating an image, removing a background, or turning a still into video. A Creative Tool lets you pick a model, fill in task-shaped inputs, and run it, while the work stays an ordinary model node in your graph with its own takes and connections. ## What is a Creative Tool? A Creative Tool is a presentation, not a separate kind of model. Each tool has a fixed job (its objective), a list of models that can do that job, a default model, and a form built from the selected model's own settings. When you run a Creative Tool, Nodespell runs the concrete model node you chose. Pricing, results, and connections are exactly those of that model node. Creative Tools appear in three places: - **Create** workspace mode, where each tool has its own composer and a **Generate**-style run button. See [Workspace modes](https://docs.nodespell.com/start/workspace-modes/). - **Graph** workspace mode, where you can add a tool to the canvas as a node and edit it in the **Parameters** panel. See [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/). - The node menu of an eligible model node, where **Show as Creative Tool** switches that node to the guided presentation. ## All Creative Tools Nodespell currently has 19 Creative Tools. The default model is the one a tool selects when you open it; you can [choose another model](https://docs.nodespell.com/creative-tools/choose-a-model/) at any time. | Tool | Starts from | Returns | Run button | Default model | | ----------------------------------------------------------------------- | ------------------------------------------------------------------ | --------------------------------- | --------------------- | ------------------------------- | | [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) | A prompt, plus reference images when the model accepts them | Image | **Generate** | GPT Image 2.5 Flare | | [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) | One or more source images and an edit instruction | Image | **Edit** | GPT Image 2.5 Flare Edit | | [Image Upscaler](https://docs.nodespell.com/creative-tools/image-upscaler/) | One image | Larger image | **Upscale** | Crystal Upscaler | | [Cinematic Shot](https://docs.nodespell.com/creative-tools/cinematic-shot/) | A prompt plus camera, lens, shot, look, and lighting choices | Image | **Generate** | GPT Image 2 | | [Variations](https://docs.nodespell.com/creative-tools/variations/) | One source image and a new shot type or written instruction | Image | **Generate** | Nano Banana 2.1 | | [Relight](https://docs.nodespell.com/creative-tools/relight/) | One source image and designed lights or a lighting reference image | Image | **Generate** | Nano Banana 2 | | [Remove Background](https://docs.nodespell.com/creative-tools/remove-background/) | One image | Image with the background removed | **Remove background** | Silhouette | | [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) | One image, or first and last frames, and a prompt | Video | **Generate** | Seedance 2.5 Image to Video | | [Text to Video](https://docs.nodespell.com/creative-tools/text-to-video/) | A prompt only | Video | **Generate** | Seedance 2.5 Text to Video | | [Reference Images to Video](https://docs.nodespell.com/creative-tools/reference-images-to-video/) | Several reference images and a prompt | Video | **Generate** | Seedance 2.5 Reference to Video | | [Modify Video](https://docs.nodespell.com/creative-tools/modify-video/) | A source video and instructions | Video | **Modify** | Gemini Omni Flash Edit | | [Extend Video](https://docs.nodespell.com/creative-tools/extend-video/) | A source clip to continue before, after, or in both directions | Video | **Extend** | Wan 3.0 Extend Video | | [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/) | One video | Higher-resolution video | **Upscale** | Topaz Video Upscaler | | [Speak](https://docs.nodespell.com/creative-tools/speak/) | A person, a script or audio, and a voice | Talking video | **Generate** | P-Video Avatar | | [Audio Generator](https://docs.nodespell.com/creative-tools/audio-generator/) | A prompt for speech, music, or sound effects | Audio | **Generate** | ElevenLabs TTS V3 | | [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/) | A script, with optional voice-reference audio | Speech audio | **Generate** | ElevenLabs TTS V3 | | [Music Generator](https://docs.nodespell.com/creative-tools/music-generator/) | A description, with optional audio references | Music | **Generate** | Google Lyria 3 Pro | | [Sound Effects Generator](https://docs.nodespell.com/creative-tools/sound-effects-generator/) | A description or an existing video | Sound effect audio | **Generate** | ElevenLabs Sound Effects V2 | | [Generate Text](https://docs.nodespell.com/creative-tools/generate-text/) | A prompt, with optional visual references | Text | **Generate** | GPT Text | In Create, the tool rail pins **Image Generator**, **Image to Video**, **Audio Generator**, and **Image Upscaler**. Open **All tools** to find the rest, grouped under headings such as **Image**, **Video**, **Audio**, and **Text**. The tool name at the top of a composer is also a menu that switches between tools of the same media type. ## How do Creative Tools relate to model nodes? Every Creative Tool run executes a concrete model node, such as Nano Banana 2 or Topaz Video Upscaler. The Creative Tool decides how that node's settings are shown: the most common controls first, the rest under **More settings**, and tool-specific controls such as Cinematic Shot's camera choices. | | Creative Tool presentation | Model node presentation | | ----------------- | ------------------------------------------------ | ---------------------------------------- | | What runs | The selected model node | The same model node | | Cost and credits | The model's own estimate | The model's own estimate | | Takes and history | Kept on the node | Kept on the node | | Connections | Unchanged | Unchanged | | Controls | Grouped, task-first, with **More settings** | Every native parameter in **Parameters** | | Model choice | Model picker limited to models that fit the tool | Fixed to that model | You can switch a node between the two presentations without changing what it does. A model that does not appear in a tool's model list is still available as an ordinary model node from the [Tab menu](https://docs.nodespell.com/graph/add-nodes/). ## Frequently asked questions ### How many Creative Tools does Nodespell have? Nodespell has 19 Creative Tools: seven for images, seven for video, four for audio, and one for text. The table on this page lists every tool with its input, result, and default model. ### Does a Creative Tool cost more than the model node it uses? No. A Creative Tool runs the selected model node, so the credit estimate and the charge are those of that model. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ### Can I use a model that a Creative Tool does not list? Yes, as an ordinary model node in Graph. Each Creative Tool only lists models whose inputs and results fit its job, so some models appear only in the Tab menu or the model hub. ### Do Creative Tools create anything hidden in my project? No. Running a Creative Tool from Create adds a visible model node to the current Level of your graph. You can open **Graph** at any time to see and edit that node. ## Related - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/): Featured models, filters, estimates, and switching safely. - [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/): Add tools as graph nodes and switch presentation. - [Make your first image](https://docs.nodespell.com/start/first-image/): A guided first run in Create. - [Takes](https://docs.nodespell.com/graph/takes/): Browse and choose the results each run keeps. --- # Choose a model for a Creative Tool > Use the Nodespell model picker and model hub to compare models, filter by maker, read estimates, and switch models without losing work. Source: https://docs.nodespell.com/creative-tools/choose-a-model/ · Last updated: 2026-11-10 The model picker is where you choose which AI model a Nodespell Creative Tool runs. The picker lists only models that fit the tool's job, shows featured models first, previews what changes before you switch, and shows a credit estimate per run on paid plans. The separate model hub in Graph adds any model to the canvas as a model node. ## Open the model picker The model picker is the model card at the top of every Creative Tool, in the Create composer and in the **Parameters** panel for a Creative Tool node on the canvas. The card shows the model maker's logo, the maker name, and the model name. 1. Open a Creative Tool, for example **Image Generator** in Create. 2. Select the model card. The picker opens with the search field focused. 3. Point at a model, or move focus to it, to see its details in the preview pane. 4. Select the model name, or select **Use** followed by the model name in the preview pane. In Create, you can step through the list without opening it: use the **Previous model** and **Next model** arrows beside the card. They are unavailable when a tool has only one model. ## What do the sections in the model picker mean? The picker list has up to three sections, in this order. | Section | What it contains | | -------------- | ------------------------------------------------------------------------------------------------------------------- | | **Favourites** | Models you marked as favourites that fit this tool. Appears only when you have one. | | **Featured** | Models Nodespell features for this kind of work. Appears only when the tool has featured models. | | **All models** | Every model that fits the tool, grouped by model family, with the maker named when it differs from the family name. | Inside **All models**, each family shows its versions as small buttons. Older versions with a newer one available are collapsed behind a **+** count; select it to show them, and **Show fewer** to hide them again. A dot marks a model that is new. When you type in the search field, the sections are replaced by a flat list of matches. You mark favourites with the star in the model hub or the Node Browser. The picker uses the same favourites. ## Filter and search models Search matches model names, makers, and features (the placeholder reads "Search models, makers, features"). Next to the search field, three menus narrow the list: | Filter | Narrows the list to | | --------------- | ------------------------------------------------------------- | | **Model maker** | One maker, such as Google or OpenAI | | **Features** | Models with one capability, such as accepting reference media | | **Best for** | Models suited to one task | Select **Clear filters** to remove all three. If nothing matches, the picker shows "No compatible results found." ## Read the model preview The preview pane shows the model you point at, or your current model when you point at nothing. | Item | What it tells you | | ------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Name and **New** badge | The model and whether it is new | | Maker | Who makes the model, plus a note such as "Nano Banana 2.1 available" when a newer version exists | | Summary | A one-paragraph description, when the pane is wide enough | | Fact badges | **Refs** (accepts reference media), **Audio** (can generate native audio), **Voice**, **Music**, or **SFX** for audio models, and output resolution, upscale, or clip-duration ranges | | **Paid plan** note | Shown for models that are part of paid plans | | **Current model** or **Switching from** | Whether this model is already selected | | **Your settings carry over** or a list of changes | What happens to your current settings if you switch | | **Cost per run** | The estimated credits per run, shown as current → new when you compare | ## How are credit estimates shown? On paid plans, the picker shows an estimate beside each model family and in the preview pane, and the run button shows the estimate for your current settings. When you set more than one generation in Create, the run button shows **Estimated total for** the number of runs. Estimates use these forms: | Display | Meaning | | ------------------------ | -------------------------------------- | | `~` followed by a number | Roughly that many credits | | A number followed by `+` | At least that many credits | | **Varies** | Cost depends on the size of your input | | **Unknown** | No estimate is available | On a free plan, a locked **Cost estimate** slot appears instead. Estimates follow your settings, so changing resolution or duration can change them. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). > **Note:** An estimate is not a fixed price. Model prices change, and the final charge depends on the completed work. ## What happens to my settings when I switch model? Nodespell moves each setting to the new model by its meaning, not by its name. A prompt stays a prompt, an aspect ratio maps to the closest one the new model supports, and an image size maps to the nearest size the new model can make. A value the new model cannot accept falls back to that model's default. What happens next depends on whether the node has run. | Situation | What Nodespell does | | ------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | A Create draft that has not run | Switches immediately. The preview pane lists any setting that changes or is **not available**. | | A newer version in the same family | Shows "_model_ available" under the card with an **Upgrade** button. In a Create draft, if settings would change, an **Upgrade** dialog asks you to review the transfer first; on the canvas, the same rules as any other switch apply. | | A canvas Creative Tool node with no takes, nothing lost | Switches in place and tells you how many settings were adjusted. | | A canvas node with no takes, something would be lost | Opens **Review what will change**, listing **Settings that won't carry over** and **Connections that won't carry over**. Select **Change model** or **Cancel**. | | A canvas node that already has takes | Opens **Create a new model branch?**. A new node gets the new model; the original node, its selected take, and everything downstream stay unchanged. Select **Create branch** or **Cancel**. | | A model node selected while you work in Create | Opens **Create a model branch?** with **Preserved** and **Not copied** counts. Select **Create branch draft** to start a draft for a new downstream node, or **Keep current model**. | Switching never moves takes from one model to another. Earlier takes stay with the node that made them. See [Branch and switch presentation](https://docs.nodespell.com/graph/branch-and-switch-presentation/). ## Browse every model in the model hub The model hub is a Graph panel for adding model nodes directly, outside any Creative Tool. Press `Shift+M` in the Graph Editor to show or hide it, and press `Escape` to close it. | Part of the model hub | What it does | | -------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------- | | **Image**, **Video**, **Audio**, **Text**, **3D** | Show models that produce that type of output | | Search | Finds models by name ("Search AI models...") | | **Featured**, **Recent**, **Favourites**, **All models** | Switch between collections; **Recent** and **Favourites** appear when they have entries | | **Filters** | **Model maker**, **Model family**, **Features**, and **Best for** | | Maker headings | Group the models under each maker | | Star | Adds or removes the model from your favourites | | Hover detail | Shows the description, input and output types, and, on paid plans, the estimate with example prices where available | Select a model card to add it to the canvas, drag it to place it, or double-click it to add it and close the hub. If the selected node's outputs do not match what a model accepts, the hover detail shows "Not compatible with current selection." ## Frequently asked questions ### Why does a model not appear in a Creative Tool's picker? Each Creative Tool lists only models whose inputs and outputs fit its job and that are available to run. Use the model hub or the Tab menu to add any other model as an ordinary model node. ### Why can't I see credit estimates in the picker? Model credit estimates are shown on paid plans. On a free plan, a locked **Cost estimate** slot appears instead; you can still run models your plan allows. ### Does switching model delete my earlier results? No. Takes stay with the node that made them. If a node already has takes, a model change creates a new branch node and leaves the original untouched. ### Where do I set favourite models? Use the star on a model card in the model hub, or **Favourite** in the Node Browser. The model picker shows the same favourites in its **Favourites** section. ## Related - [Creative Tools overview](https://docs.nodespell.com/creative-tools/) — every tool and its default model - [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/) — switching models on a graph node - [Branch and switch presentation](https://docs.nodespell.com/graph/branch-and-switch-presentation/) — how branches protect earlier work - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) — what estimates mean for your balance --- # Use Creative Tools as nodes on the canvas > Add Nodespell Creative Tools to the graph canvas, edit them in Parameters, and switch any eligible node between Creative Tool and model node views. Source: https://docs.nodespell.com/creative-tools/on-the-canvas/ · Last updated: 2026-11-10 A Creative Tool on the canvas is an ordinary Nodespell model node shown with the guided controls of a Creative Tool, such as Image Generator or Relight. You add it to the graph without picking a model first, edit it in the **Parameters** panel, connect it like any node, and switch it back to the plain model node view whenever you want. ## Add a Creative Tool to the graph A new Creative Tool node starts with the tool's default model and is named after the tool, for example "Image Generator". Nodespell selects the new node so you can fill in its inputs straight away. | Where | How | | ----------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Tab menu | Press `Tab`, then search for the tool name. Creative Tools are listed under **Creative tools**, and they appear first when you browse a media category. | | Tab menu from a socket | Drag from an output socket and release to open the menu. Only tools whose model can take that connection are listed, and the new node is connected for you. | | Canvas right-click menu | Right-click empty canvas, open a media section such as **Image**, then a task. The first entry, **Creative tool** with **Any model** beside it, adds the Creative Tool; the entries below it add specific model nodes. | Tools without a matching task row in the right-click menu, such as Cinematic Shot or Relight, appear as their own entries in their media section. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/) for the Tab menu in general. ## Edit a Creative Tool node in Parameters Select a Creative Tool node and open the **Parameters** panel. Instead of the native parameter list, the panel shows: 1. The tool name, with the [model picker](https://docs.nodespell.com/creative-tools/choose-a-model/) card below it. 2. The tool's main controls, grouped into sections such as **References**, **Output**, and **Settings**. Specialist tools add their own controls here, such as Cinematic Shot's camera and shot design choices or Relight's lighting stage. 3. **More settings**, which holds every remaining parameter of the model and shows how many you changed. 4. A run button labelled with the tool's action, such as **Generate**, **Edit**, or **Upscale**, with the credit estimate beside it on paid plans. The run button appears only when this node is the only selected node. It runs the node the same way [Run Selected Tasks](https://docs.nodespell.com/graph/run-nodes/) does, including any upstream work that needs to run first. Each run adds one take to the node; to make several takes in one go, use the run options described in [Run nodes](https://docs.nodespell.com/graph/run-nodes/). When a node's input is connected, the matching control shows that it is controlled by the connected node. A connected prompt shows "Prompt controlled by an input node" with a **Disconnect** button. ## Switch between Creative Tool and model node view The presentation of a node is a display choice. Switching it never changes the model, parameter values, connections, takes, position, or name. | To switch | Do this | | -------------------------- | -------------------------------------------------------------------------------------------------------------------------------------- | | Creative Tool → model node | Right-click the node and select **Show as** _model name_ **node**, for example **Show as Nano Banana 2 node**. | | Model node → Creative Tool | Right-click an eligible model node and select **Show as Creative Tool**. If the model fits several tools, choose one from the submenu. | **Show as Creative Tool** appears only for models that a Creative Tool lists. For Cinematic Shot, Variations, and Relight, the option appears only when the node already holds that tool's settings, for example a node you first set up as a Relight tool. ## What happens when I change the model on a canvas node? Choose another model from the picker in **Parameters**. Nodespell plans the change before applying it. | Node state | Result | | ----------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | No takes yet, and every setting and connection fits the new model | The model changes in place. A notice tells you how many settings were adjusted or could not carry over. | | No takes yet, but some settings or connections would be lost | **Review what will change** lists **Settings that won't carry over** and **Connections that won't carry over**, each marked **Incoming** or **Outgoing**. Select **Change model** or **Cancel**. | | The node already has takes | **Create a new model branch?** explains that a new node keeps the original model and every previous result intact. Select **Create branch** to add the new node, or **Cancel**. | If the graph changes while the dialog is open, Nodespell refreshes the review and asks you to confirm again. Cancelling leaves the node exactly as it was. See [Branch and switch presentation](https://docs.nodespell.com/graph/branch-and-switch-presentation/). ## Save a default model for a Creative Tool Set up a Creative Tool node the way you like, right-click it, and choose **Save as my default…**. The dialog explains that the current model becomes the default model for that tool's nodes you add from now on, in any project. Existing nodes do not change. See [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/). ## How is a canvas Creative Tool different from Create? Both use the same tool definitions, model lists, controls, and estimates. | | Create workspace | Creative Tool node in Graph | | --------------------------------- | -------------------------------------------------- | ------------------------------ | | Where you edit | The Create composer | The **Parameters** panel | | Generations per run | **Generations** stepper, 1 to 4 | One take per run | | Adds to the graph | A new model node at the current Level when you run | The node you placed | | Model switch on a node with takes | **Create a model branch?** draft | **Create a new model branch?** | ## Frequently asked questions ### Does showing a node as a Creative Tool change how it runs? No. The node runs the same model with the same parameters and connections. Only the controls in **Parameters** change. ### Why is Show as Creative Tool missing from a node's menu? The node's model is not listed by any Creative Tool, or it fits only Cinematic Shot, Variations, or Relight and the node does not hold that tool's settings. The node still works as a model node. ### Can I rename a Creative Tool node? Yes. A custom node name is kept when you change the model or switch presentation. ### Why is there no run button in Parameters? The run button appears only when the Creative Tool node is the only selected node. Select just that node, or use [Run Selected Tasks](https://docs.nodespell.com/graph/run-nodes/). ## Related - [Creative Tools overview](https://docs.nodespell.com/creative-tools/) — every tool and what it returns - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — the picker, estimates, and switching - [Add nodes](https://docs.nodespell.com/graph/add-nodes/) — the Tab menu and Node Browser - [Parameters](https://docs.nodespell.com/graph/parameters/) — the panel that hosts Creative Tool controls - [Takes](https://docs.nodespell.com/graph/takes/) — results kept on each node --- # Generate images with Image Generator > Use the Nodespell Image Generator Creative Tool to create new images from a prompt, choose a model, set size and quality, and make up to four takes. Source: https://docs.nodespell.com/creative-tools/image-generator/ · Last updated: 2026-11-10 Image Generator is the Nodespell Creative Tool for creating a new image from a written prompt. You describe the image, choose a model, set its size and other options, and select **Generate**. Some models also accept reference images to guide subject or style. The result is an image take on a model node in your graph. ## When should I use Image Generator? Use Image Generator when you start from words, not from an existing picture. | If you want to | Use | | -------------------------------------------------------------------------- | ------------------------------------------------- | | Create a new image from a description | Image Generator | | Control camera, lens, shot type, color look, and lighting as named choices | [Cinematic Shot](https://docs.nodespell.com/creative-tools/cinematic-shot/) | | Change an existing image with an instruction | [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) | | Reframe an existing image from a new angle | [Variations](https://docs.nodespell.com/creative-tools/variations/) | | Turn a still into motion | [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) | ## Generate an image 1. Open **Create** and select **Image Generator** in the tool rail. 2. Check the model card at the top. The default model is GPT Image 2.5 Flare; select the card to [choose another model](https://docs.nodespell.com/creative-tools/choose-a-model/). 3. Write your prompt. 4. If the model has a **References** section, add reference images with **Choose assets**. You can pick from your Assets or upload files. 5. Set the size and any options under **Output** and **Settings**. Open **More settings** for the rest. 6. Set **Generations** to the number of images you want, from 1 to 4. 7. Select **Generate**. On a Graph canvas, add Image Generator from the `Tab` menu and use the same controls in **Parameters**. See [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/). ## Image Generator inputs and controls Image Generator has no required media input. Only the prompt is required for most models. The controls fixed by the tool itself are: | Control | What it does | Options or default | | ----------------------------------- | ------------------------------------------------------------------- | --------------------------------------- | | Model card | Chooses the model that runs | Default GPT Image 2.5 Flare | | **Previous model** / **Next model** | Steps through the tool's models without opening the picker | Create only | | **Prompt** | Describes the image you want | Required for most models | | **References** | Reference images that guide the result, when the model accepts them | Model-dependent | | **More settings** | Every other setting the model has, with a count of changed values | Collapsed | | **Generations** | How many runs to submit at once | 1 to 4, default 1 (Create only) | | **Generate** | Validates the inputs and runs the model | Shows the credit estimate on paid plans | ## Which models work with Image Generator? Image Generator lists every available model that creates images from a prompt. The picker shows these under **Featured**: - GPT Image 2.5 Flare (default) - GPT Image 2.5 Sunburst - Nano Banana 2.1 - Grok Imagine Image 2.0 - FLUX 3 Image - Recraft V4.1 **All models** lists the rest, including Nano Banana 2, Nano Banana Pro, Seedream 5 Pro, GPT Image 2, Ideogram, Recraft, and Qwen models. Model availability can change; the picker always shows the current list. ## How do the controls change with the model? The settings below **Prompt** come from the selected model. With the default model, GPT Image 2.5 Flare, you see: | Setting | Options (GPT Image 2.5 Flare) | | ------------------------------------ | --------------------------------------------------------------------------------------------------------------- | | **Image Size** | Auto, 1024x1024, 1536x1024, 1024x1536, 2048x2048, 2048x1152, 2560x1440, 3840x2160, or Custom; default 1024x1024 | | **Custom Width** / **Custom Height** | 256 to 3840 pixels, for Custom size | | **Quality** | Low, Medium, High, XHigh, Max, or Auto; default High | | **Output Format** | PNG, JPEG, or WebP; default PNG | | **Compression** | 0 to 100, default 90 | GPT Image 2.5 Flare takes a prompt only. Models such as Nano Banana 2 and Nano Banana 2.1 add **Reference images**, **Aspect ratio**, and **Resolution**; others add settings such as a seed or quality level. When you switch model, Nodespell carries your prompt and the closest matching size to the new model and lists anything that changes. ## Where does the generated image go? Selecting **Generate** adds a model node to the current Level of your graph and runs it once per generation. Each finished image is kept as a [take](https://docs.nodespell.com/graph/takes/) on that node. The **Results** area of the composer shows the current result with **Show older result** and **Show newer result**, **View fullscreen**, and **Continue with this result**. **Continue with this result** opens a choice of **Image Upscaler** (a larger, refined version) or **Image to Video** (use the image as the starting frame). Nodespell keeps the exact image as a visible Asset node and connects it to the next tool. ## Common problems | Problem | What to do | | ------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------- | | "This value is required." under the prompt | Write a prompt before you select **Generate**. | | No **References** section | The selected model does not accept reference images. Choose a model with the **Refs** badge in the picker. | | "Some generations could not be submitted." | Part of the batch failed to start. The runs that started are kept in the node's history; select **Generate** again for more. | | "The Create action could not be completed." | Nothing ran. Check your inputs and try again. | | The estimate shows **Varies** | The cost depends on your input or output size; the run button updates as you change settings. | ## Frequently asked questions ### Which model does Image Generator use by default? Image Generator starts with GPT Image 2.5 Flare. You can switch to any model the picker lists, and your prompt carries over. ### How many images can I generate at once? In Create, set **Generations** from 1 to 4. Each generation is a separate run of the same node, and each result is kept as a take. On the canvas, each run makes one take. ### Can I use reference images with Image Generator? Yes, when the selected model accepts them. Models such as Nano Banana 2.1 show a **Reference images** control; the default model takes a prompt only. ### Does Image Generator edit an existing image? No. To change an existing image, use [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) or [Variations](https://docs.nodespell.com/creative-tools/variations/). ## Related - [Make your first image](https://docs.nodespell.com/start/first-image/) — a guided first run - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — featured models, estimates, switching - [Cinematic Shot](https://docs.nodespell.com/creative-tools/cinematic-shot/) — camera-led image generation - [Takes](https://docs.nodespell.com/graph/takes/) — browse and choose results - [Image Upscaler](https://docs.nodespell.com/creative-tools/image-upscaler/) — enlarge a generated image --- # Edit an image with Image Editor > Use the Nodespell Image Editor Creative Tool to change existing images with a written instruction, combine several sources, and limit edits with a mask. Source: https://docs.nodespell.com/creative-tools/image-editor/ · Last updated: 2026-11-10 Image Editor is the Nodespell Creative Tool for changing an existing image with a written instruction. You give it one or more source images, describe what should change, and select **Edit**. Depending on the model, you can also combine several images, add references, or limit the change with a mask. The result is a new image take; your source stays unchanged. ## When should I use Image Editor? Use Image Editor when you have a picture and want a specific change to it. | If you want to | Use | | ----------------------------------------------------------- | ------------------------------------------------------- | | Change, add, or remove something in an image by instruction | Image Editor | | Combine elements from several images into one | Image Editor, with a model that accepts several images | | See the same subject from a new framing or angle | [Variations](https://docs.nodespell.com/creative-tools/variations/) | | Change only the lighting | [Relight](https://docs.nodespell.com/creative-tools/relight/) | | Cut the subject out | [Remove Background](https://docs.nodespell.com/creative-tools/remove-background/) | | Make a new image from words only | [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) | ## Edit an image 1. Open **Create**, select **All tools**, and choose **Image Editor**. You can also switch to it from the tool-name menu while another image tool is open. 2. Check the model card. The default model is GPT Image 2.5 Flare Edit. 3. Add your source image with **Choose assets**: pick it from your Assets or upload a file. 4. Write the instruction in the prompt, for example what to change and what to keep. 5. Add a mask or references if the model offers them, and review **Output** and **More settings**. 6. Set **Generations** from 1 to 4. 7. Select **Edit**. To edit an image already in your graph, add Image Editor on the canvas and connect the image's output to the source input. See [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/). ## Image Editor inputs and controls Image Editor always requires a source image. The tool marks the model's first image input as required, so **Edit** stays blocked until it has an image or a connection. | Control | What it does | Options or default | | ----------------- | --------------------------------- | --------------------------------------------------------------------------------- | | Model card | Chooses the editing model | Default GPT Image 2.5 Flare Edit | | Source image | The image to change | Required; the label follows the model, for example **Images** or **Source image** | | **Prompt** | The edit instruction | Required for most models | | **More settings** | Every other setting the model has | Collapsed | | **Generations** | Number of runs to submit | 1 to 4, default 1 (Create only) | | **Edit** | Runs the model | Shows the credit estimate on paid plans | ## Which models work with Image Editor? Image Editor lists available models that take an image and return an edited image. The picker shows these under **Featured**: - GPT Image 2.5 Flare Edit (default) - GPT Image 2.5 Sunburst Edit - Nano Banana 2.1 - Grok Imagine Image 2.0 - FLUX 3 Image **All models** includes Nano Banana 2, Nano Banana Pro, Nano Banana 2 Lite, Seedream 5 Pro, GPT Image 2 Edit, Flux Kontext Dev, Flux Fill Pro, Ideogram 4.5 Edit, Qwen Image Edit models, and others. ## How do the controls change with the model? With the default model, GPT Image 2.5 Flare Edit, Image Editor shows: | Setting | Options (GPT Image 2.5 Flare Edit) | | ----------------- | ---------------------------------------------------------------------------- | | **Images** | 1 to 16 source or reference images to edit or combine; required | | **Mask** | Optional image that marks the editable area of the first image | | **Image Size** | Auto, preset sizes from 1024x1024 to 3840x2160, or Custom; default 1024x1024 | | **Quality** | Low, Medium, High, XHigh, Max, or Auto; default High | | **Output Format** | PNG, JPEG, or WebP; default PNG | | **Compression** | 0 to 100, default 90 | | **Background** | Auto or Opaque; default Auto | Other models differ. Nano Banana 2 and Nano Banana 2.1 show a **Source image** input plus **Aspect ratio** and **Resolution**. Some models accept extra **Reference images** that guide the edit without being the image you change. To build a mask in Nodespell, see [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/). ## Where does the edited image go? Selecting **Edit** adds a model node to the current Level of your graph and runs it. Each edited image is kept as a [take](https://docs.nodespell.com/graph/takes/) on that node, and the **Results** area shows it with **Show older result**, **Show newer result**, and **View fullscreen**. The source image is not modified. ## Common problems | Problem | What to do | | -------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------- | | **Edit** does nothing and the source shows "This value is required." | Add a source image or connect one. | | "Open Graph to edit this media input." | This input cannot be edited in Create. Open **Graph** and edit the node there. | | The change spreads beyond the area you wanted | Use a model with a mask input, or describe what must stay the same. | | The result has a different size or shape | Set **Image Size** or **Aspect ratio**; some models default to their own size rather than the source's. | ## Frequently asked questions ### Can Image Editor combine several images? Yes, with a model that accepts several images. GPT Image 2.5 Flare Edit takes 1 to 16 images in its **Images** input. ### Does Image Editor overwrite my original image? No. The edit is a new take on a new model node. Your source image and its node stay unchanged. ### What is the difference between Image Editor and Variations? Image Editor applies your instruction to the image. [Variations](https://docs.nodespell.com/creative-tools/variations/) is built for reframing one image as a new shot type or a directed variation, with a shot-type picker. ### Can I edit only part of an image? Yes, when the model has a **Mask** input. The mask marks the area the model may change. ## Related - [Variations](https://docs.nodespell.com/creative-tools/variations/) — new framings of one image - [Relight](https://docs.nodespell.com/creative-tools/relight/) — change lighting only - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare editing models - [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/) — make masks for local edits - [Recipe: revise without starting over](https://docs.nodespell.com/recipes/revise-without-starting-over/) — iterate on a result --- # Enlarge images with Image Upscaler > Use the Nodespell Image Upscaler Creative Tool to enlarge an image, set a scale factor, and continue from a generated result to a sharper version. Source: https://docs.nodespell.com/creative-tools/image-upscaler/ · Last updated: 2026-11-10 Image Upscaler is the Nodespell Creative Tool for making an image larger while improving its detail. You give it one image, choose a model and a scale factor, and select **Upscale**. The result is a new, larger image take; the source image is not changed. You can upscale any image from your Assets or continue directly from an image result. ## When should I use Image Upscaler? Use Image Upscaler when the picture is right but too small or too soft for its final use. | If you want to | Use | | ----------------------------------------- | ------------------------------------------------------------------------------ | | Increase the resolution of a still image | Image Upscaler | | Increase the resolution of a video | [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/) | | Change content, not just size | [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) | | Generate at a higher size from the start | [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) with a larger size setting | | Resize without AI, to an exact pixel size | A Resize node; see [Transforms and crop](https://docs.nodespell.com/edit/transforms-and-crop/) | ## Upscale an image 1. Open **Create** and select **Image Upscaler** in the tool rail. 2. Check the model card. The default model is Crystal Upscaler. 3. Add the image with **Choose assets**, from your Assets or by uploading a file. 4. Set **Scale factor** and any other options the model shows. 5. Set **Generations** from 1 to 4 if you want several attempts. 6. Select **Upscale**. ## Continue from a result After any image result in Create, select **Continue with this result** and choose **Image Upscaler**. Nodespell keeps that exact result as a visible Asset node named "Selected historical result", connects it to the upscaler's source input, and opens the Image Upscaler composer. A notice says "This tool will use the exact selected result through a visible Asset node." Select **Remove** to drop that source and pick another. Nothing runs until you select **Upscale**. ## Image Upscaler inputs and controls | Control | What it does | Options or default | | ----------------- | ------------------------------- | --------------------------------------- | | Model card | Chooses the upscaling model | Default Crystal Upscaler | | **Source image** | The image to enlarge | Required, one image | | **More settings** | Any other setting the model has | Collapsed when present | | **Generations** | Number of runs to submit | 1 to 4, default 1 (Create only) | | **Upscale** | Runs the model | Shows the credit estimate on paid plans | ## Which models work with Image Upscaler? The picker shows these under **Featured**: - Crystal Upscaler (default), tuned for portraits and faces - Clarity Pro Upscaler - Google Upscaler **All models** also lists Bria Increase Resolution, P Image Upscale, Qwen Image Edit Upscale, Recraft Crisp Upscale, and SeedVR Upscale Image. ## How do the controls change with the model? With Crystal Upscaler, the main control is **Scale factor**, from 1 to 6 with a default of 2. A scale factor of 2 doubles the width and height. Other upscalers use their own controls, such as an upscale factor or a target size. The picker's fact badges show each model's upscale range before you switch. Credit estimates for upscalers often depend on the size of the input image, so the estimate may read **Varies** until the image is known. ## Where does the upscaled image go? Selecting **Upscale** adds a model node to the current Level of your graph and runs it. The larger image is kept as a [take](https://docs.nodespell.com/graph/takes/) on that node and shown in **Results**. Use **View fullscreen** to check detail at full size. ## Common problems | Problem | What to do | | --------------------------------------------- | -------------------------------------------------------------------------------------------------- | | "This value is required." on **Source image** | Add an image or connect one before you select **Upscale**. | | Added detail looks wrong | AI upscalers add detail that was not in the source. Try a lower scale factor or a different model. | | "The saved source is no longer available." | The result you continued from is gone. Select **Remove** and choose another image. | ## Frequently asked questions ### How much can Image Upscaler enlarge an image? It depends on the model. Crystal Upscaler accepts a scale factor from 1 to 6. Check the upscale range badge in the model picker for other models. ### Does upscaling change my original image? No. The upscaled image is a new take on a new node. The source stays as it was. ### Can I upscale a video with Image Upscaler? No. Image Upscaler accepts still images only. Use [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/) for video. ### Can I upscale an image I generated earlier, not the latest one? Yes. Show the result you want in **Results**, then select **Continue with this result**. Nodespell uses that exact take. ## Related - [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) — create the image first - [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/) — the video equivalent - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare upscalers - [Takes](https://docs.nodespell.com/graph/takes/) — keep and choose results - [Export](https://docs.nodespell.com/edit/export/) — get the final image out --- # Compose a still with Cinematic Shot > Use the Nodespell Cinematic Shot Creative Tool to generate a still with named camera, lens, film stock, shot type, color look, and lighting choices. Source: https://docs.nodespell.com/creative-tools/cinematic-shot/ · Last updated: 2026-11-10 Cinematic Shot is the Nodespell Creative Tool for generating a still image with the vocabulary of a camera department. You write the scene, then pick a camera character, lens, focal length, aperture, film stock, shot type, color look, and lighting from named lists with example images. Nodespell turns those choices into instructions for the selected image model and returns an image take. ## When should I use Cinematic Shot? Use Cinematic Shot when the look of the frame matters as much as its content and you would rather pick from named options than write photographic language yourself. | If you want to | Use | | ---------------------------------------------------------------- | --------------------------------------------------- | | Set camera, lens, shot size, color look, and lighting as choices | Cinematic Shot | | Generate from a prompt with only the model's own settings | [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) | | Reframe an existing image as a new shot type | [Variations](https://docs.nodespell.com/creative-tools/variations/) | | Change lighting on an image you already have | [Relight](https://docs.nodespell.com/creative-tools/relight/) | ## Create a cinematic still 1. Open **Create**, select **All tools**, and choose **Cinematic Shot**. 2. Write the scene in the prompt: subject, setting, action. 3. Select the **Camera** button to open the camera dialog. Choose **Camera character**, **Lens character**, **Focal length**, **Aperture**, and **Film stock**. Point at an option to see an example image. 4. Under **Shot design**, choose **Shot type**, **Color look**, and **Lighting**. 5. Open **More cinematic controls** for **Motion blur**, **Grain**, **Halation**, and **Tonal response**. 6. Check the model card and the model's own settings, such as size, under **Output** and **More settings**. 7. Set **Generations** from 1 to 4, then select **Generate**. Every Cinematic Shot choice starts at **Standard**, which adds nothing for that setting. Change only the ones you care about. ## Cinematic Shot controls These controls belong to the tool and stay the same whichever model you choose. | Control | Section | What it sets | Options | | -------------------- | --------------------------- | ---------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Camera character** | **Camera** | The rendering character of a camera body | **Standard** or 15 named cameras, from ARRI ALEXA 35 and Sony VENICE 2 to Kodak Super 8, Holga 120N, and Fisher-Price PXL-2000 | | **Lens character** | **Camera** | Lens rendering, flare, and falloff | **Standard** or 17 lenses, such as Cooke S4/i, ARRI Master Anamorphic, Helios 44-2 58mm, and Lensbaby Sweet 50 | | **Focal length** | **Camera** | Field of view and perspective | **Standard**, 8mm fisheye, 14mm, 24mm, 35mm, 50mm, 85mm, 135mm, 200mm, 300mm, 600mm | | **Aperture** | **Camera** | Depth of field | **Standard**, f/1.2, f/1.4, f/2, f/2.8, f/4, f/5.6, f/8, f/11, f/16 | | **Film stock** | **Camera** | Color and texture of a film stock | **Standard** or 21 stocks in five groups: Current professional cinema, Legacy professional cinema, Professional & editorial photography, Iconic & enthusiast, Black and white | | **Shot type** | **Shot design** | Framing and camera position | **Standard** or 23 options grouped as Shot size, Composition, Camera angle, and Viewpoint | | **Color look** | **Shot design** | Overall grade | **Standard** or 17 looks grouped as Natural & restrained, Stylized palette, and Photochemical & period | | **Lighting** | **Shot design** | Lighting style | **Standard** or 30 setups grouped as Natural, available & practical; Direction & texture; Contrast & mood; and Portrait patterns | | **Motion blur** | **More cinematic controls** | Movement in the frame | **Standard**, Frozen action, Natural motion blur, Strong motion blur, Long-exposure trails | | **Grain** | **More cinematic controls** | Film grain | **Standard**, Clean / nearly grainless, 65mm ultra-fine grain, 35mm fine grain, 35mm pronounced grain, Super 16 grain, 16mm coarse grain, Super 8 heavy grain, Pushed-process grain | | **Halation** | **More cinematic controls** | Glow around bright edges | **Standard**, None, Subtle edge halation, Classic red-orange halation, Warm amber halation, Broad soft halation, Dense crimson halation, Neon highlight halation | | **Tonal response** | **More cinematic controls** | Contrast and curve shape | **Standard**, Low contrast, Soft film contrast, Balanced contrast, Punchy contrast, Hard contrast, Lifted blacks, Soft highlight roll-off | Shot type options include Extreme close-up, Close-up, Medium shot, Cowboy shot, Wide shot, Establishing shot, Insert shot, Low angle, Overhead, Worm's-eye view, Dutch angle, Three-quarter view, Over-the-shoulder, and Point-of-view (POV). Equipment names are descriptive creative references; Nodespell is not affiliated with or endorsed by their owners. ## Which models work with Cinematic Shot? Cinematic Shot uses a fixed list of image models. It has no **Featured** section, so the picker opens on **All models**. | Model | Notes | | ------------------- | ------------------------ | | GPT Image 2 | Default | | Nano Banana 2 | Accepts reference images | | Nano Banana Pro | | | Seedream 5 Pro | | | Seedream 4.5 | | | Recraft V4.1 Pro | | | Ideogram V4 | | | Flux 2 Dev | | | RunwayML Gen4 Image | | ## How do the results change with the model? The Cinematic Shot choices are the same for every model, but each model interprets them in its own way, so the same settings give different images on different models. The rest of the form comes from the model. With GPT Image 2 you also set **Resolution** (1024 x 768 by default, up to 3840 x 2160, or Custom), **Output format** (PNG, JPEG, or WEBP), and **Quality** (Low, Medium, or High). Nano Banana 2 instead offers **Reference images**, **Aspect ratio**, and **Resolution**. > **Tip:** Named equipment and film stocks guide the model's style. They do not simulate the real optics or film chemistry exactly. ## Where does the image go? Selecting **Generate** adds a model node to the current Level of your graph and runs it. Each image is kept as a [take](https://docs.nodespell.com/graph/takes/) on that node and appears in **Results**. Nodespell stores your Cinematic Shot choices with the node, so selecting the node later brings them back in the composer. On the canvas, a Cinematic Shot node keeps the same controls in **Parameters**; see [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/). ## Common problems | Problem | What to do | | ------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------- | | A choice seems to have no effect | Some models follow style cues more loosely. Try another model or strengthen the prompt. | | The picker shows no **Featured** section | Expected: Cinematic Shot lists all its models under **All models**. | | **Show as Creative Tool** does not offer Cinematic Shot for a model node | Cinematic Shot is offered only for nodes that already hold Cinematic Shot settings. Add a new Cinematic Shot from the `Tab` menu instead. | ## Frequently asked questions ### What does Standard mean in Cinematic Shot? **Standard** leaves that aspect to the prompt and the model. Every Cinematic Shot control starts at **Standard**. ### Can I use reference images with Cinematic Shot? Yes, with a model that accepts them, such as Nano Banana 2. The default model, GPT Image 2, takes a prompt only. ### Does Cinematic Shot make video? No. Cinematic Shot returns a still image. To animate it, continue with [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/). ### Are the camera and lens names real equipment? They are names of real cameras, lenses, and film stocks, used as creative references for the model. Nodespell is not affiliated with their makers. ## Related - [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) — prompt-only image generation - [Variations](https://docs.nodespell.com/creative-tools/variations/) — reframe an existing image by shot type - [Relight](https://docs.nodespell.com/creative-tools/relight/) — redesign lighting on an image - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare the supported models - [Recipe: product shot](https://docs.nodespell.com/recipes/product-shot/) — a worked example --- # Reframe an image with Variations > Use the Nodespell Variations Creative Tool to reframe one image as a new shot type or angle, or to direct a custom variation with your own words. Source: https://docs.nodespell.com/creative-tools/variations/ · Last updated: 2026-11-10 Variations is the Nodespell Creative Tool for making a new version of one existing image. In **Reframe** mode you choose a shot type, such as a close-up or a low angle, and the model redraws the scene from that framing while keeping the subject recognizable. In **Custom** mode you write exactly what should change and what should stay. Each run returns a new image take. ## When should I use Variations? Use Variations when you like an image and want another shot of the same subject or scene. | If you want to | Use | | ------------------------------------------------------------ | ----------------------------------------------------------------------------- | | See the same subject from a new framing, angle, or viewpoint | Variations, **Reframe** mode | | Make a directed change while keeping the subject consistent | Variations, **Custom** mode, or [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) | | Combine several images or edit with a mask | [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) | | Generate a fresh frame with camera and lens choices | [Cinematic Shot](https://docs.nodespell.com/creative-tools/cinematic-shot/) | | Change only the lighting | [Relight](https://docs.nodespell.com/creative-tools/relight/) | ## Reframe an image 1. Open **Create**, select **All tools**, and choose **Variations**. 2. Add the image under **Source image** with **Choose assets**. Variations uses exactly one source image. 3. Leave **Variation mode** on **Reframe**. 4. Open **Shot type** and choose a framing. Point at or focus an option to preview it. 5. Check the model card and the output settings. 6. Set **Generations** from 1 to 4, then select **Generate**. To make a directed change instead, set **Variation mode** to **Custom** and describe the change in **Variation instructions**. ## Variations controls | Control | What it does | Options or default | | -------------------------- | ----------------------------------------------------------- | ----------------------------------------------------- | | **Source image** | The image to transform | Required, one image | | **Variation mode** | Chooses how the variation is described | **Reframe** (default) or **Custom** | | **Shot type** | The new framing in **Reframe** mode | 23 options in four groups; default Three-quarter view | | **Variation instructions** | What should change and what should stay, in **Custom** mode | Required in **Custom** mode | | Model card | Chooses the model | Default Nano Banana 2.1 | | **More settings** | The model's remaining settings | Collapsed | | **Generations** | Number of runs to submit | 1 to 4, default 1 (Create only) | | **Generate** | Runs the model | Shows the credit estimate on paid plans | **Shot type** groups and options: | Group | Options | | ------------ | -------------------------------------------------------------------------------------------------------------- | | Shot size | Extreme close-up, Close-up, Medium close-up, Medium shot, Cowboy shot, Full shot, Wide shot, Extreme wide shot | | Composition | Establishing shot, Insert shot | | Camera angle | Eye level, Low angle, High angle, Overhead, Aerial, Worm's-eye view, Dutch angle | | Viewpoint | Front view, Three-quarter view, Profile view, Back view, Over-the-shoulder, Point-of-view (POV) | In **Reframe** mode, Nodespell tells the model to create one reframed variation at the chosen shot type, using the source as the visual reference and keeping its subject, scene, and recognizable details where possible. In **Custom** mode, your instruction is sent as written. ## Which models work with Variations? Variations uses a fixed list of image models. The picker shows Nano Banana 2.1 (the default) and FLUX 3 Image under **Featured**, and lists all of these under **All models**: - Nano Banana 2.1, Nano Banana 2, Nano Banana Pro, Nano Banana 2 Lite - GPT Image 2 Edit - Seedream 5 Pro - Flux Kontext Dev, FLUX 3 Image - Qwen Edit Multiangle, Qwen Image 3, Qwen Image 3 Pro - Ideogram 4.5 Edit ## How do the results change with the model? Each model follows the shot type in its own way, and some keep the subject closer to the source than others. Qwen Edit Multiangle also receives matching camera moves for the chosen shot type, such as moving closer for a close-up or tilting for a low or high angle. The output controls come from the model: Nano Banana 2.1, for example, adds **Aspect ratio** and **Resolution**, while other models offer their own size or quality settings. When you switch model, your source image and settings carry over where the new model supports them. ## Where does the variation go? Selecting **Generate** adds a model node to the current Level of your graph and runs it. Each variation is kept as a [take](https://docs.nodespell.com/graph/takes/) on that node and shown in **Results**. Nodespell stores the mode and shot type with the node, so selecting it later brings them back. The source image is not changed. ## Common problems | Problem | What to do | | ---------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------- | | "Describe the variation you want." | You are in **Custom** mode with empty **Variation instructions**. Write an instruction or switch to **Reframe**. | | The subject changes too much | Try another model, or use **Custom** mode and state what must stay the same. | | The prompt shows "Prompt controlled by an input node" instead of the mode controls | A node is connected to the prompt input. Select **Disconnect** to use **Variation mode** and **Shot type** again. | ## Frequently asked questions ### Can Variations use more than one source image? No. Variations works from exactly one source image. Some models also accept separate reference images; to combine several images, use [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/). ### What is the default shot type in Variations? Three-quarter view, in **Reframe** mode. Change it from the **Shot type** list. ### Do I need to write a prompt in Reframe mode? No. In **Reframe** mode the shot type is enough. **Variation instructions** are required only in **Custom** mode. ### Is Variations the same as Image Editor? They overlap, but Variations is built around one source image and a shot-type choice, with a fixed list of models suited to reframing. Image Editor follows a free-form instruction and can take several images. ## Related - [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) — instruction-based edits - [Cinematic Shot](https://docs.nodespell.com/creative-tools/cinematic-shot/) — the same shot types for new images - [Recipe: consistent character](https://docs.nodespell.com/recipes/consistent-character/) — keep a subject consistent across shots - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare variation models - [Takes](https://docs.nodespell.com/graph/takes/) — compare and keep variations --- # Change lighting with Relight > Use the Nodespell Relight Creative Tool to relight an image with key, fill, and rim lights on a 3D stage, or match the lighting of a reference image. Source: https://docs.nodespell.com/creative-tools/relight/ · Last updated: 2026-11-10 Relight is the Nodespell Creative Tool for changing the lighting of an existing image while keeping it recognizable. You either design up to three lights (key, fill, and rim) on an interactive 3D stage with exposure, softness, and color, or you give Relight a second image whose lighting it should match. Each run returns a new image take. ## When should I use Relight? Use Relight when the content of an image is right and only the light needs to change. | If you want to | Use | | ------------------------------------------------- | ------------------------------------------------------------- | | Place and shape lights yourself | Relight, **Design lighting** | | Copy the lighting of another image | Relight, **Match reference** | | Change content, not just light | [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) | | Set a lighting style while generating a new image | [Cinematic Shot](https://docs.nodespell.com/creative-tools/cinematic-shot/) | | Adjust exposure and color without AI | A color node; see [Color correction](https://docs.nodespell.com/edit/color-correction/) | ## Design new lighting 1. Open **Create**, select **All tools**, and choose **Relight**. 2. Add the image under **Source image** with **Choose assets**. 3. Leave **Relight method** on **Design lighting**. 4. Choose a **Lighting preset**, such as **Golden hour** or **Product softbox**. The preset sets up one to three lights. 5. On the **Lighting stage**, select **Key**, **Fill**, or **Rim** to pick a light, then drag it to a new position and adjust **Exposure**, **Softness**, and **Light color**. 6. Open **Lighting notes** to add mood, contrast, shadow, or preservation notes if needed. 7. Set **Generations** from 1 to 4, then select **Generate**. ## Match the lighting of a reference image 1. Add the **Source image**. 2. Set **Relight method** to **Match reference**. 3. Under **Lighting reference**, choose the image whose lighting should guide the result. 4. Choose **Reference influence**: **Subtle**, **Balanced**, or **Strong**. 5. Select **Generate**. **Match reference** works only with models that accept two images. If the current model cannot, the method shows "Choose a reference-capable model to match a lighting reference." ## Relight controls | Control | What it does | Options or default | | ----------------------- | ------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Source image** | The image to relight | Required, one image | | **Relight method** | Designed lights or a reference image | **Design lighting** (default) or **Match reference** | | **Lighting preset** | Starting light setup | Soft studio (default, 1 light), Natural window (2), Golden hour (2), Dramatic side light (1), Product softbox (2), Rim light (3), Neon contrast (3), Moonlight (2), Custom | | **Light role** | Selects the light you are editing | **Key**, **Fill**, **Rim** | | Add light | Adds the next light: **Fill light**, then **Rim light** | Up to three lights; switches the preset to **Custom** | | **Remove light** | Removes the selected light; removing the fill light also removes the rim light | The key light cannot be removed | | **Exposure** | Brightness of the selected light | −3.0 to +3.0 EV in 0.1 steps; key light default 0.0 EV | | **Softness** | Shadow edge, from hard to soft | 0 to 100%; key light default 45% | | **Light color** | **Natural** white light or **Color light** | **Natural** uses **Color temperature**, 2,000 to 12,000 K in 100 K steps, key light default 5,600 K. **Color light** offers Red, Orange, Green, Cyan, Blue, Magenta, or a custom color | | **Lighting notes** | Extra guidance not covered by the controls | Optional | | **Lighting reference** | Image to copy lighting from, in **Match reference** | Required in that method | | **Reference influence** | How strongly the reference applies | **Subtle**, **Balanced** (default), **Strong** | Editing any light, or adding or removing one, switches **Lighting preset** to **Custom**. Double-click the **Exposure**, **Softness**, or **Color temperature** slider to reset it. ## Use the lighting stage The **Lighting stage** is an interactive 3D view of your lights around the subject. | Action | Result | | -------------------------- | ----------------------------------- | | Drag the control sphere | Rotates the selected light freely | | Drag the X, Y, or Z ring | Rotates around that world axis | | Drag the outer yellow ring | Rotates around the view axis | | Click a light | Selects it | | Arrow keys | Rotate around X and Y | | `Q` and `E` | Rotate around Z | | **Reset camera view** | Returns the stage view to its start | All stage gestures use the primary button, so they work on a trackpad. If the browser cannot show 3D, the stage reads "The 3D lighting preview is unavailable in this browser." ## Which models work with Relight? Relight uses a fixed list of image models and has no **Featured** section. | Model | **Design lighting** | **Match reference** | | ----------------------- | ------------------- | ------------------- | | Nano Banana 2 (default) | Yes | Yes | | Nano Banana 2 Lite | Yes | Yes | | Nano Banana Pro | Yes | Yes | | Seedream 5 Pro | Yes | Yes | | GPT Image 2 Edit | Yes | Yes | | Flux Kontext Dev | Yes | No | In Create, when **Match reference** is selected, the picker lists only the models that support it. On the canvas, **Match reference** is unavailable while the node uses Flux Kontext Dev. ## Where does the relit image go? Selecting **Generate** adds a model node to the current Level and runs it. Each result is kept as a [take](https://docs.nodespell.com/graph/takes/) and shown in **Results**. Nodespell stores the full lighting setup with the node, so selecting it later restores the method, preset, and lights. If a result holds lighting settings the node no longer has, Relight offers **Restore to node**, which writes them back without running anything. ## Common problems | Problem | What to do | | --------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------ | | "Choose a lighting reference image." | Add an image under **Lighting reference**, or switch to **Design lighting**. | | "Choose a reference-capable model to match a lighting reference." | Pick a model marked Yes for **Match reference** above. | | "Match reference needs two local image slots. Disconnect the source input before generating." | The source comes from a connection. Disconnect it and choose the source image in the tool. | | The subject changes more than the light | Add a preservation note in **Lighting notes**, such as what must stay unchanged, or try another model. | ## Frequently asked questions ### How many lights can Relight use? Up to three: a key light, a fill light, and a rim light. The key light is always present. ### Is the Relight 3D stage a real render? No. The stage shows where your lights are. The AI model reads the setup as instructions and paints the new lighting; it does not ray-trace the scene. ### Can I relight using another photo's lighting? Yes. Set **Relight method** to **Match reference**, add the photo under **Lighting reference**, and choose a **Reference influence**. ### Does Relight change my original image? No. The relit image is a new take on a new node. The source stays unchanged. ## Related - [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) — instruction-based edits - [Cinematic Shot](https://docs.nodespell.com/creative-tools/cinematic-shot/) — lighting styles for new images - [Color correction](https://docs.nodespell.com/edit/color-correction/) — non-AI exposure and color - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — Relight model differences - [Takes](https://docs.nodespell.com/graph/takes/) — compare lighting versions --- # Cut out a subject with Remove Background > Use the Nodespell Remove Background Creative Tool to isolate an image subject as a transparent cutout, with edge controls for hair and soft edges. Source: https://docs.nodespell.com/creative-tools/remove-background/ · Last updated: 2026-11-10 Remove Background is the Nodespell Creative Tool for isolating the subject of an image and removing everything behind it. You give it one image and select **Remove background**. With the default model, Silhouette, the result is a transparent cutout at the original image size, with options for hair, fur, and soft edges. The cutout is a new image take you can composite or export. ## When should I use Remove Background? Use Remove Background when you need the subject on its own, for a product shot, a composite, or a new background. | If you want to | Use | | ------------------------------------------------------ | --------------------------------------------------------------------------- | | Remove the background from a still image automatically | Remove Background | | Pull a key from a green or blue screen | A Keyer node; see [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/) | | Replace the background with a described scene | [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) | | Place the cutout over a new background | Remove Background, then a compositor; see [Compositing](https://docs.nodespell.com/edit/compositing/) | ## Remove a background 1. Open **Create**, select **All tools**, and choose **Remove Background**. 2. Add the image under the source input with **Choose assets**. 3. Check the model card. The default model is Silhouette. 4. Open **More settings** to change the method or edges if needed. 5. Select **Remove background**. On a Graph canvas, add Remove Background from the `Tab` menu or from the **Image** section of the canvas right-click menu, and connect an image to it. See [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/). ## Remove Background inputs and controls | Control | What it does | Options or default | | --------------------- | ------------------------ | --------------------------------------- | | Source image | The image to cut out | Required, one image | | Model card | Chooses the model | Default Silhouette | | **More settings** | The model's own options | Collapsed | | **Generations** | Number of runs to submit | 1 to 4, default 1 (Create only) | | **Remove background** | Runs the model | Shows the credit estimate on paid plans | ## Which models work with Remove Background? The picker shows Silhouette (the default) and Bria Remove Background under **Featured**. **All models** also lists Remove Background Carve Tracer and Transparent Background. ## How do the controls change with the model? With Silhouette, you get these settings. The last three sit inside the **Edge adjustments** group: | Setting | What it does | Options or default | | ----------------------- | -------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------ | | **Method** | How edges are treated | **General** (default) for most subjects, or **Hair & soft edges** for hair, fur, and partly transparent boundaries | | **Expand / contract** | Grows the cutout to restore clipped edges, or shrinks it to remove a thin fringe | −64 to 64 output-image pixels, default 0 | | **Feather** | Softens the cutout boundary | 0 to 64 pixels radius, default 0 | | **Remove color fringe** | Reduces background color bleeding into soft edges; may change edge colors | Off by default | Silhouette returns a cutout at the original image dimensions and re-estimates any transparency already in the input. Other models have their own settings. Transparent Background, for example, has **Format**, **Reverse** (remove the foreground instead), **Threshold**, and **Background Type**. ## Where does the cutout go? Selecting **Remove background** adds a model node to the current Level of your graph and runs it. The cutout is kept as a [take](https://docs.nodespell.com/graph/takes/) on that node and shown in **Results**. Connect the node's output to a compositor to place the subject on a new background, or [export](https://docs.nodespell.com/edit/export/) it. Keep a format with transparency, such as PNG, when you export. ## Common problems | Problem | What to do | | --------------------------------------------- | ------------------------------------------------------------------------------------------------------------------ | | Hair or fur is cut off | Set **Method** to **Hair & soft edges**, or use **Expand / contract** with a positive value. | | A thin line of old background around the edge | Use a small negative **Expand / contract**, or turn on **Remove color fringe**. | | The edge looks too sharp | Add a little **Feather**. | | Transparent or glass subjects lose detail | Difficult backgrounds and transparent subjects may still need manual cleanup, for example with masks in the graph. | ## Frequently asked questions ### Does Remove Background keep the original image size? With Silhouette, yes: the cutout matches the original image dimensions. Other models may differ. ### Can Remove Background work on video? No. Remove Background accepts still images only. ### What format is the cutout? The cutout has a transparent background. Keep it in a format that supports transparency, such as PNG, when you export. ### Can I fine-tune the edge after the cut? Yes. Change **Expand / contract**, **Feather**, or **Remove color fringe** and run again, or refine the matte with [masks and keying](https://docs.nodespell.com/edit/masks-and-keying/) nodes. ## Related - [Compositing](https://docs.nodespell.com/edit/compositing/) — place the cutout on a new background - [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/) — refine edges and mattes - [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) — replace a background by instruction - [Recipe: product shot](https://docs.nodespell.com/recipes/product-shot/) — a worked example - [Export](https://docs.nodespell.com/edit/export/) — save the cutout --- # Animate an image with Image to Video > Use the Nodespell Image to Video Creative Tool to animate one still or guide a shot between a start frame and an end frame. Source: https://docs.nodespell.com/creative-tools/image-to-video/ · Last updated: 2026-11-10 Image to Video is the Nodespell Creative Tool for turning one still image into a video clip. You supply a start image and a prompt that describes the motion, and many models also accept an optional end image so you can guide the shot from a first frame to a last frame. ## When to use Image to Video Choose Image to Video when one picture must anchor the shot: the subject, composition and look of your image become the opening frame of the clip. | If you have… | Use | | ------------------------------------------------------------------------- | ----------------------------------------------------------------------- | | One image that should become the first frame | **Image to Video** | | A first image and a last image to move between | **Image to Video** with a model that has an end-frame field | | Several images of a character, product or style, but no fixed first frame | [Reference Images to Video](https://docs.nodespell.com/creative-tools/reference-images-to-video/) | | Only a written idea | [Text to Video](https://docs.nodespell.com/creative-tools/text-to-video/) | | An existing clip you want to make longer | [Extend Video](https://docs.nodespell.com/creative-tools/extend-video/) | | An existing clip you want to change | [Modify Video](https://docs.nodespell.com/creative-tools/modify-video/) | ## Generate a video from an image 1. Switch the workspace to **Create** and select **Image to Video** in the tool rail. Image to Video has its own rail button. 2. Check the model at the top of the form. The default is **Seedance 2.5 Image to Video**. Use **Previous model** and **Next model** to step through models, or open the picker to browse them. 3. Add your still to the start-frame field. Select the field to open **Choose assets**, where you pick from your Assets or upload a file. 4. If the model shows an end-frame field and you want the shot to finish on a specific picture, add the last image there. 5. Write a prompt that describes the motion, camera direction and, for models that generate sound, the audio you want. 6. Set duration, resolution and any other options the model shows. 7. Set **Generations** to the number of clips you want, from 1 to 4. 8. Select **Generate**. Create needs a window at least 1080 pixels wide. In a narrower window, use **Open Graph** and work with the model node directly. ## Use a first frame and a last frame First/last-frame generation means the model starts the clip on one image and ends it on another, inventing the motion between them. It is available only on models that expose an end-frame input; Image to Video does not add one to models that lack it. - With a model that has curated Create labels, the fields read **Start frame** and **End frame**. - Other models show their own input names, for example **Start Image** and **End Image** on Seedance 2.5 Image to Video, or **Image** and **Last Frame** on Veo 3.1. - On most models the end frame is optional. Some first/last-frame models, such as FLUX.3 First/Last Frame to Video, require both images. Keep both images at the same framing and aspect ratio, and describe in the prompt how the camera or subject should travel between them. ## Image to Video controls Every Creative Tool shares the same frame around the model's own settings. These controls belong to Image to Video itself: | Control | What it does | Options and defaults | | ----------------------------------- | ------------------------------------------------------------------------------- | ------------------------------------ | | Tool switcher (title) | Switches to another video Creative Tool and keeps you in Create | All video tools | | Model picker | Chooses the model; the picker opens on **Featured** when featured models exist | Default: Seedance 2.5 Image to Video | | **Previous model** / **Next model** | Steps through the eligible models in order | — | | Start-frame field | The source image; required on most models | One image | | **More settings** | Shows less-used model settings and how many you changed | Collapsed | | **Generations** | How many clips one select of **Generate** submits | 1 to 4, default 1 | | **Generate** | Validates the form, adds or updates the node in your graph and submits the runs | Shows the credit estimate | The fields between the model picker and **More settings** come from the selected model. They are grouped under **Prompt**, **References**, **Output** and **Settings**. ## Recommended models for Image to Video The **Featured** collection for Image to Video currently lists, in order: - Seedance 2.5 Image to Video (default) - MiniMax H3 Image to Video - Gemini Omni Flash Image To Video - Veo 3.1 - Kling V3 Turbo Pro Image To Video **All models** shows every model that accepts one image as its main input. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) for filters, favorites and switching. ## What changes with the model? Duration, resolution, aspect ratio, audio and frame inputs all come from the model, so the form changes when you switch. - **Duration:** ranges differ widely. Seedance 2.5 Image to Video offers `auto` or 4 to 30 seconds; other models offer fixed choices such as 5 or 10 seconds. - **Resolution:** Seedance 2.5 offers 480p, 720p and 1080p (default 720p); some models go to 4K, and some have no resolution setting. - **Audio:** models with a **Generate Audio** toggle can create synchronized sound with the picture. Seedance 2.5 has it on by default. - **Framing:** on Seedance 2.5 Image to Video the output framing follows the supplied image. - **Draft Mode:** Seedance 2.5 offers **Draft Mode**, a 480p preview. You can render a draft at 1080p within seven days with Seedance 2.5 Draft to Final in [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/). Check the credit estimate on **Generate** after changing duration or resolution, since both usually change the cost. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Start from an image you generated in Create In an image Creative Tool such as [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/), select **Continue with this result** on an image in **Results**, then choose **Image to Video**. Nodespell keeps that exact image as a visible Asset node and uses it as the starting frame. The notice "This tool will use the exact selected result through a visible Asset node." appears, with **Remove** to drop the source. Nothing runs until you select **Generate**. ## Where does the video go? **Generate** places an Image to Video model node in your current graph level and submits the run. Each finished clip appears in **Results** in the Create form, where you can step with **Show older result** and **Show newer result** or **View fullscreen**. The node keeps every clip as part of its history; see [Takes](https://docs.nodespell.com/graph/takes/). Generated media also appears under **Generated** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). To work with the same node in Graph, see [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/). ## Common problems - **A field shows "This value is required." after you select Generate.** The start image or the prompt is missing. - **"Prompt controlled by an input node"** — a node in the graph feeds the prompt. Edit that node, or select **Disconnect** to type here. - **"The saved source is no longer available."** The continued result or Asset was removed. Select **Remove** and add a new image. - **The clip ends somewhere unexpected.** The model has no end-frame input, or the field is empty. Pick a model with an end-frame field. ## Frequently asked questions ### Can Image to Video use a first and last frame? Yes, when the selected model has an end-frame input, such as Seedance 2.5 Image to Video, Veo 3.1 or Kling 4.0 Image to Video. Models without that input animate from the start image only. ### What is the default Image to Video model? The default is Seedance 2.5 Image to Video. It accepts a start image, an optional end image, durations of `auto` or 4 to 30 seconds, and 480p to 1080p output. ### How many clips can I generate at once? Set **Generations** from 1 to 4. Each generation is a separate run with its own credit cost, and the estimate on **Generate** shows the total. ### Does Image to Video add sound? Only when the model supports it. Models with a **Generate Audio** toggle, including Seedance 2.5 Image to Video and Veo 3.1, can generate synchronized audio. ## Related - [Make your first image-to-video result](https://docs.nodespell.com/start/first-image-to-video/) — a guided first run - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — picker, featured models and switching - [Reference Images to Video](https://docs.nodespell.com/creative-tools/reference-images-to-video/) — guide a shot with several images - [Extend Video](https://docs.nodespell.com/creative-tools/extend-video/) — make a finished clip longer - [Takes](https://docs.nodespell.com/graph/takes/) — browse and choose results on the node - [Image to video shot recipe](https://docs.nodespell.com/recipes/image-to-video-shot/) — an end-to-end example --- # Generate a video from a prompt with Text to Video > Use the Nodespell Text to Video Creative Tool to create a video clip from a written prompt alone, with no source image or video. Source: https://docs.nodespell.com/creative-tools/text-to-video/ · Last updated: 2026-11-10 Text to Video is the Nodespell Creative Tool for making a video clip from a written prompt alone. It lists only models that take text and no source media, so the whole shot—subject, setting, camera and, on some models, sound—comes from your description. ## When to use Text to Video Choose Text to Video when you have no picture to start from, or when you want the model to design the shot freely. | If you have… | Use | | ------------------------------------------------ | ---------------------------------------------------------------------------------------------- | | Only a written idea | **Text to Video** | | One image that should be the first frame | [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) | | Several images for a character, product or style | [Reference Images to Video](https://docs.nodespell.com/creative-tools/reference-images-to-video/) | | An existing clip to change or lengthen | [Modify Video](https://docs.nodespell.com/creative-tools/modify-video/) or [Extend Video](https://docs.nodespell.com/creative-tools/extend-video/) | If you need the same character in several shots, a text prompt alone rarely keeps it consistent. Generate a still first and use Image to Video, or use Reference Images to Video. ## Generate a video from text 1. Switch the workspace to **Create**. Select **Image to Video** in the tool rail, then open the tool switcher at the top of the form and choose **Text to Video**. You can also find it under **All tools**. 2. Check the model. The default is **Seedance 2.5 Text to Video**. 3. Write the prompt: who or what is in the shot, where it is, how the camera moves, the mood, and any sound you want. 4. Set duration, resolution and aspect ratio if the model shows them. 5. Set **Generations** from 1 to 4. 6. Select **Generate**. ## Text to Video controls These controls belong to the Creative Tool. The prompt, duration, resolution and the rest come from the selected model. | Control | What it does | Options and defaults | | ----------------------------------- | -------------------------------------------------------- | ----------------------------------- | | Tool switcher (title) | Moves to another video Creative Tool | All video tools | | Model picker | Chooses the model | Default: Seedance 2.5 Text to Video | | **Previous model** / **Next model** | Steps through eligible models | — | | **Prompt** | The description of the shot; required on most models | Empty | | **More settings** | Less-used model settings, with a count of changed values | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Generate** | Adds or updates the model node and submits the runs | Shows the credit estimate | ## Recommended models for Text to Video The **Featured** collection for Text to Video lists, in order: - Seedance 2.5 Text to Video (default) - MiniMax H3 Text to Video - Gemini Omni Flash - FLUX.3 Text to Video - Kling V3 Turbo Pro Text To Video Open **All models** in the picker to see the rest. [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) explains favorites, filters and switching. ## What changes with the model? - **Duration:** Seedance 2.5 Text to Video offers `auto` or 4 to 30 seconds. Others offer fixed steps, for example 5 or 10 seconds. - **Resolution:** Seedance 2.5 offers 480p, 720p and 1080p, default 720p. Some models offer up to 4K; some have no resolution control. - **Aspect ratio:** Seedance 2.5 offers `auto`, 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16. Aspect-ratio choices appear as one-tap chips when a model has a short list. - **Audio:** Seedance 2.5 Text to Video generates synchronized audio unless you turn off **Generate Audio**. Many models produce silent video. - **Negative prompt:** some models show a **Negative prompt** field for things to avoid. - **Draft Mode:** Seedance 2.5 offers a 480p **Draft Mode** preview you can later render at 1080p with Seedance 2.5 Draft to Final in [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/), within seven days. Higher resolution and longer duration usually raise the estimate shown on **Generate**. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Write a prompt that works for video - Describe motion explicitly: "the camera pushes in slowly", "she turns toward the window". - Name the shot size and angle: wide shot, close-up, low angle. - Keep one main action per clip. Long chains of events are better as separate shots joined in the [Timeline](https://docs.nodespell.com/edit/timeline/). - For models that generate audio, say what should be heard: dialogue, ambience, music. Reusable prompt text can live in your Saved text library; see [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/). ## Where does the video go? **Generate** creates a Text to Video model node in the current graph level and runs it. Finished clips appear in **Results** in the Create form, with **Show older result**, **Show newer result** and **View fullscreen**. The node keeps its history of results; see [Takes](https://docs.nodespell.com/graph/takes/). Generated media also appears under **Generated** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ## Common problems - **"This value is required."** The prompt is empty. - **The prompt field is read-only with "Prompt controlled by an input node".** A connected node supplies the prompt. Edit that node or select **Disconnect**. - **A selected node shows "This node uses Graph controls".** That node's model is not one of the reviewed models for this Creative Tool. Select **Open in Graph** to edit it there, or **Start new**. ## Frequently asked questions ### What is the default Text to Video model? The default is Seedance 2.5 Text to Video. It supports `auto` or 4 to 30 second clips, 480p to 1080p output, several aspect ratios and optional synchronized audio. ### Can I add an image to Text to Video? No. Text to Video lists only models without media inputs. To start from an image, use [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) or [Reference Images to Video](https://docs.nodespell.com/creative-tools/reference-images-to-video/). ### How long can a Text to Video clip be? It depends on the model. Seedance 2.5 Text to Video and Wan 3.0 Text to Video go up to 30 seconds; many others stop at 10 or 15 seconds. ### Will the same prompt give the same video twice? Usually not. Some models expose a **Seed** setting under **More settings**, which can make results more repeatable, but no model guarantees identical output. ## Related - [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) — start from a still - [Reference Images to Video](https://docs.nodespell.com/creative-tools/reference-images-to-video/) — keep a subject consistent - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare and switch models - [Run nodes](https://docs.nodespell.com/graph/run-nodes/) — what happens when a node runs - [Compare models side by side](https://docs.nodespell.com/recipes/compare-models-side-by-side/) — try one prompt on several models --- # Guide a video with Reference Images to Video > Use the Nodespell Reference Images to Video Creative Tool to generate a clip guided by several images of a subject, product or style. Source: https://docs.nodespell.com/creative-tools/reference-images-to-video/ · Last updated: 2026-11-10 Reference Images to Video is the Nodespell Creative Tool for generating a video clip guided by several images at once. The images describe who or what should appear, the style, or the composition, and the prompt describes the shot; unlike Image to Video, no single image has to become the first frame. ## When to use Reference Images to Video | If you have… | Use | | --------------------------------------------------------- | ------------------------------------------------- | | Several images of a character, product, outfit or setting | **Reference Images to Video** | | One image that must be the opening frame | [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) | | No images, only a prompt | [Text to Video](https://docs.nodespell.com/creative-tools/text-to-video/) | | A finished clip you want to restyle or edit | [Modify Video](https://docs.nodespell.com/creative-tools/modify-video/) | Reference Images to Video lists models whose image input accepts many images. Many of them also accept reference videos and reference audio. ## Generate a video from reference images 1. Switch the workspace to **Create**. Select **Image to Video** in the tool rail, open the tool switcher at the top of the form, and choose **Reference Images to Video**. It is also listed under **All tools**. 2. Check the model. The default is **Seedance 2.5 Reference to Video**. 3. Add images to the reference-images field (labelled **Reference images**, or **Reference Images** on models that use their own labels). Select the field to open **Choose assets** and pick several Assets or upload files. 4. Add reference videos or reference audio if the model offers those fields and you need motion or timing guidance. 5. Write a prompt that says what each reference is for, for example "the woman in `@Image1` wears the jacket from `@Image2`". 6. Set duration, resolution and aspect ratio. 7. Set **Generations** from 1 to 4, then select **Generate**. ## Refer to images in the prompt Several models number references so the prompt can point at them. Seedance 2.5 Reference to Video, for example, reads `@Image1`, `@Image2`, `@Video1` and `@Audio1` in the order you added the files. Reorder references if the numbering matters, and name each one's role in the prompt. Models that do not use this syntax treat the images as general guidance. ## Reference Images to Video controls | Control | What it does | Options and defaults | | ----------------------------------- | --------------------------------------------------- | ------------------------------------------- | | Tool switcher (title) | Moves to another video Creative Tool | All video tools | | Model picker | Chooses the model | Default: Seedance 2.5 Reference to Video | | **Previous model** / **Next model** | Steps through eligible models | — | | Reference-images field | The ordered set of guide images | Multiple images; limits depend on the model | | **More settings** | Less-used model settings | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Generate** | Adds or updates the model node and submits the runs | Shows the credit estimate | Reference video, reference audio, duration, resolution, aspect ratio and audio fields appear only when the model has them, under the model's own labels. ## Recommended models for Reference Images to Video The **Featured** collection lists, in order: - Seedance 2.5 Reference to Video (default) - MiniMax H3 Reference to Video - Veo 3.1 Veo 3.1 appears in both Image to Video and Reference Images to Video because it accepts a start frame, a last frame and reference images. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) for the rest of the list. ## What changes with the model? - **Reference types:** Seedance 2.5 Reference to Video accepts reference images, videos and audio. Its audio references need at least one image or video reference. Other models accept images only. - **Required references:** some models, such as Gemini Omni Flash Reference To Video, require at least one reference image; on Seedance 2.5 all reference inputs are optional. - **Duration:** Seedance 2.5 offers `auto` or 4 to 30 seconds; MiniMax H3 offers 5 to 15 seconds. - **Resolution and audio:** Seedance 2.5 offers 480p to 1080p and a **Generate Audio** toggle that is on by default. - **Start frame:** a few reference models, such as Kling 4.0 Reference to Video, also take an optional start image alongside the references. ## Get consistent subjects - Use clean, well-lit images that show the subject clearly. Several angles of the same subject help more than many near-duplicates. - Keep the references to what the shot needs. Unrelated images pull the result in different directions. - State in the prompt what must stay the same, such as face, logo or outfit. For a full walkthrough, see [Keep a character consistent](https://docs.nodespell.com/recipes/consistent-character/). ## Where does the video go? **Generate** adds a Reference Images to Video model node to the current graph level and runs it. Finished clips appear in **Results** in the Create form and on the node; see [Takes](https://docs.nodespell.com/graph/takes/). Generated media also appears under **Generated** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ## Common problems - **"This value is required."** The prompt is empty, or the model requires a reference image. - **A reference field says "Controlled by an upstream connection".** A node in the graph feeds that input. Change it in the graph, or disconnect it. - **The subject drifts between shots.** Add clearer references, refer to each image in the prompt, and keep the same model for every shot. ## Frequently asked questions ### How many reference images can I add? It depends on the model, and the field accepts only what the model allows. Read the model's description in the picker before relying on a large set. ### Is a reference image used as the first frame? No. Reference images guide the subject and style; they are not placed as the opening frame. To fix the first frame, use [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/). ### Can Reference Images to Video use video and audio references? Yes, on models that offer them. Seedance 2.5 Reference to Video, Seedance 2.0 Reference To Video and Wan 3.0 Reference to Video accept images, videos and audio. ### What is the default Reference Images to Video model? The default is Seedance 2.5 Reference to Video, which accepts mixed image, video and audio references and clips of `auto` or 4 to 30 seconds. ## Related - [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) — anchor the shot on one image - [Keep a character consistent](https://docs.nodespell.com/recipes/consistent-character/) — references across several shots - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare reference support - [Upload and organize Assets](https://docs.nodespell.com/assets/upload-and-organize/) — prepare reference images - [Connections](https://docs.nodespell.com/graph/connections/) — feed references from other nodes --- # Change an existing clip with Modify Video > Use the Nodespell Modify Video Creative Tool to edit, restyle or transform an existing video clip with a written instruction and optional references. Source: https://docs.nodespell.com/creative-tools/modify-video/ · Last updated: 2026-11-10 Modify Video is the Nodespell Creative Tool for changing a video you already have. You supply a source clip and an instruction—change the lighting, swap an object, restyle the shot—and the selected model returns a new video based on the original. Some models also accept reference images or audio. ## When to use Modify Video | You want to… | Use | | ---------------------------------------------------------- | --------------------------------------------------------------------- | | Change what is in a clip, or how it looks | **Modify Video** | | Add footage before or after a clip | [Extend Video](https://docs.nodespell.com/creative-tools/extend-video/) | | Make a clip sharper or larger without changing its content | [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/) | | Make a person in a clip speak new audio | [Speak](https://docs.nodespell.com/creative-tools/speak/) | | Trim, cut or arrange clips | [Timeline](https://docs.nodespell.com/edit/timeline/) and [Video editing](https://docs.nodespell.com/edit/video-editing/) | Modify Video does not list models whose job is extending a clip; those are in Extend Video. ## Modify a video 1. Switch the workspace to **Create**. Select **Image to Video** in the tool rail, open the tool switcher at the top of the form, and choose **Modify Video**. It is also under **All tools**. 2. Check the model. The default is **Gemini Omni Flash Edit**. 3. Add the source clip to the video field. Select it to open **Choose assets** and pick an Asset or upload a file. 4. Write the instruction. Name the change directly, for example "make it night, keep everything else the same". 5. Add reference images, keyframes or audio if the model offers them. 6. Set **Generations** from 1 to 4. 7. Select **Modify**. The primary button reads **Modify** in this tool, not **Generate**. ## Modify Video controls | Control | What it does | Options and defaults | | ----------------------------------- | --------------------------------------------------- | ------------------------------- | | Tool switcher (title) | Moves to another video Creative Tool | All video tools | | Model picker | Chooses the model | Default: Gemini Omni Flash Edit | | **Previous model** / **Next model** | Steps through eligible models | — | | Source video field | The clip to change; always required in this tool | One video | | **More settings** | Less-used model settings | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Modify** | Adds or updates the model node and submits the runs | Shows the credit estimate | The source field is labelled **Source video** on models with curated Create labels, and **Video** or **Source Video** on others. ## Recommended models for Modify Video The **Featured** collection lists: - Gemini Omni Flash Edit (default) - Runway Aleph 2 **All models** includes other editing, restyling and motion models, for example Wan 3.0 Video Edit, Kling 4.0 Video Edit, Luma Ray 3.2 Video To Video and Kling V3 Motion Control. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). ## What changes with the model? Modify Video covers several kinds of change, and each model handles a different mix: | Model | What it adds to the form | | ----------------------- | -------------------------------------------------------------------------------------------------------------------- | | Gemini Omni Flash Edit | Source video and an instruction only | | Runway Aleph 2 | Up to 5 optional keyframe images, with **Keyframe Positions** (`first`, `last`, or a time in seconds) and a **Seed** | | Wan 3.0 Video Edit | Reference images, reference audio, duration, resolution and **Generate audio** | | Kling V3 Motion Control | A required image and a required video, to drive the image with the clip's motion | - **Clip length:** limits come from the model. Runway Aleph 2, for example, accepts source videos between 2 and 30 seconds. - **Prompting:** the Gemini Omni Flash Edit description recommends short, single-change instructions and adding "Keep everything else the same". Do complex changes as separate passes. - **Voice:** the Gemini Omni Flash Edit description notes that voice editing is not supported. For new speech on a face, use [Speak](https://docs.nodespell.com/creative-tools/speak/). Longer and higher-resolution clips usually cost more; check the estimate on **Modify** and see [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Where does the modified video go? **Modify** places the model node in the current graph level and runs it. Results appear in **Results** in the Create form and on the node; see [Takes](https://docs.nodespell.com/graph/takes/). The source clip is not changed. Generated media also appears under **Generated** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ## Common problems - **"This value is required."** The source video or the instruction is missing. - **"Open Graph to edit this media input."** This media input cannot be edited in Create. Use **Open unrestricted Graph editing** and edit the node in Graph. - **The model rejects the clip.** The source is outside the model's length, size or format limits. Read the model's input description, trim the clip in the [Timeline](https://docs.nodespell.com/edit/timeline/), and try again. - **Uploaded-video editing is unavailable.** The Gemini Omni Flash Edit description notes that this depends on regional policy. Try another model. ## Frequently asked questions ### Does Modify Video overwrite my original clip? No. Modify Video creates a new result on its own node. The source Asset or node output stays as it was. ### Can Modify Video make a clip longer? No. Use [Extend Video](https://docs.nodespell.com/creative-tools/extend-video/), which lists models built to add footage before or after a clip. ### What is the default Modify Video model? The default is Gemini Omni Flash Edit. It takes a source video and a short natural-language instruction. ### Why did my old Modify Video draft open in Extend Video? Drafts and canvas cards that used an extension model now open in Extend Video, because extension moved into its own Creative Tool. Your settings are kept. ## Related - [Extend Video](https://docs.nodespell.com/creative-tools/extend-video/) — add footage to a clip - [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/) — raise resolution without changing content - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare editing models - [Video editing](https://docs.nodespell.com/edit/video-editing/) — trim and arrange clips without AI - [Revise without starting over](https://docs.nodespell.com/recipes/revise-without-starting-over/) — iterate on a result --- # Make a clip longer with Extend Video > Use the Nodespell Extend Video Creative Tool to continue a source clip before it, after it, or at both ends, with an optional prompt. Source: https://docs.nodespell.com/creative-tools/extend-video/ · Last updated: 2026-11-10 Extend Video is the Nodespell Creative Tool for continuing a clip you already have. You supply a source video and, optionally, a prompt for the new footage, and the model generates more of the shot before it, after it, or—on some models—at both ends. Most models return the whole extended clip, not only the new part. ## When to use Extend Video | You want to… | Use | | ----------------------------------------------------------- | ------------------------------------------------- | | Continue a shot past its last frame, or lead into its first | **Extend Video** | | Change what happens inside the existing clip | [Modify Video](https://docs.nodespell.com/creative-tools/modify-video/) | | Start a new shot from a still | [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) | | Join separate clips one after another | [Timeline](https://docs.nodespell.com/edit/timeline/) | Extend Video keeps the source's framing and look as far as the model can. It does not guarantee that the original frames are preserved exactly. ## Extend a clip 1. Switch the workspace to **Create**. Select **Image to Video** in the tool rail, open the tool switcher at the top of the form, and choose **Extend Video**. It is also under **All tools**. 2. Check the model. The default is **Wan 3.0 Extend Video**. 3. Add the clip to **Source video**. Select the field to open **Choose assets** and pick an Asset or upload a file. 4. Choose where new footage goes, if the model offers a direction setting. 5. Set the length. On Wan 3.0 this is **Total duration**, the length of the whole result including the source. 6. Optionally write a prompt for what happens in the new footage. On Wan 3.0 you can leave it blank for a natural continuation. 7. Set **Generations** from 1 to 4, then select **Extend**. ## Extend Video controls | Control | What it does | Options and defaults | | ----------------------------------- | --------------------------------------------------- | ----------------------------- | | Tool switcher (title) | Moves to another video Creative Tool | All video tools | | Model picker | Chooses the extension model | Default: Wan 3.0 Extend Video | | **Previous model** / **Next model** | Steps through the extension models | — | | Source video field | The clip to continue; always required | One video | | **More settings** | Less-used model settings | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Extend** | Adds or updates the model node and submits the runs | Shows the credit estimate | ## Wan 3.0 Extend Video settings Wan 3.0 Extend Video is the default model. Wan 3.0 Prime Extend Video has the same controls and generates faster at a higher cost. The model describes itself as in preview, so results can vary between runs. | Setting | What it does | Default | | ----------------------- | ------------------------------------------------------------------------------- | ---------------- | | **Source video** | One MP4 or MOV clip, 1–15 seconds, at least 16 fps, at most 100 MB | Required | | **Prompt** | Instructions for the new footage; blank gives a natural continuation | Empty | | **Extension Direction** | **Add to End**, **Add to Start** or **Both Ends** | **Add to End** | | **Total duration** | Length of the full output including the source, `Auto` or 2–30 seconds | `Auto` | | **Resolution** | 480p, 720p or 1080p | 720p | | **Reference images** | Up to 10 ordered images, referred to as `@Image1`, `@Image2` | Optional | | **Reference audio** | Up to 5 WAV or MP3 clips, 15 seconds combined at most, referred to as `@Audio1` | Optional | | **Generate Audio** | Includes generated dialogue, music or sound effects | On | | **Enhance Prompt** | Expands your prompt with extra detail | On | | **Seed**, **Watermark** | Repeatability and a visible watermark | -1 (random), off | **Total duration** must be longer than the source, and the source plus output cannot exceed 30 seconds. `Auto` needs a source shorter than 15 seconds. ## Models for Extend Video Extend Video has no featured list, so the picker opens on **All models**. The models are: | Model | Direction | Length setting | Notes | | ------------------------------- | ------------------ | ------------------------------ | ------------------------------------------------------------------- | | Wan 3.0 Extend Video (default) | End, start or both | Total output, up to 30 s | Optional prompt, image and audio references | | Wan 3.0 Prime Extend Video | End, start or both | Total output, up to 30 s | Same controls, faster | | FLUX.3 Extend Video | After the source | Continuation, `auto` or 5–20 s | Source MP4 under 50 MB and under 15 s; up to 1080p; prompt required | | Grok Imagine Video Extension | After the source | **Extension Duration**, 2–10 s | Prompt required; returns the source and extension as one clip | | Lightricks LTX 2.3 Extend Video | Start or end | Extension, up to 20 s | Optional prompt and context length | The length setting means different things: on Wan 3.0 it is the total result, on the others it is the added footage. Longer source clips can raise the cost on models that bill for the whole stitched clip; check the estimate on **Extend**. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Where does the extended video go? **Extend** places the model node in the current graph level and runs it. The result appears in **Results** in the Create form and on the node; see [Takes](https://docs.nodespell.com/graph/takes/). Your source clip is not changed. Generated media also appears under **Generated** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). To keep extending, use the new result as the next source. ## Common problems - **"This value is required."** There is no source video, or the model requires a prompt. - **The run fails on Wan 3.0 with a duration error.** **Total duration** is not longer than the source, or source plus output exceeds 30 seconds. Raise or lower **Total duration**, or trim the source. - **The source is rejected.** It is outside the model's limits for length, size, frame rate or format. Trim it in the [Timeline](https://docs.nodespell.com/edit/timeline/) and export again. - **The join is visible.** Describe continuous motion in the prompt, keep the same lighting description, and try another generation. ## Frequently asked questions ### Does Extend Video return only the new footage? On Wan 3.0 Extend Video and Grok Imagine Video Extension the result is the full extended clip, source included. Check the model description for others before cutting it into a sequence. ### Can I add footage before the start of a clip? Yes, with Wan 3.0 Extend Video or Wan 3.0 Prime Extend Video (**Add to Start** or **Both Ends**), or Lightricks LTX 2.3 Extend Video (start mode). ### How long can an extended clip be? On Wan 3.0 Extend Video the source plus output cannot exceed 30 seconds. Other models limit the added footage instead, for example 20 seconds on FLUX.3 Extend Video. ### Is a prompt required? Not on Wan 3.0 Extend Video, where a blank prompt gives a natural continuation. FLUX.3 Extend Video and Grok Imagine Video Extension require one. ### Where did extension go in Modify Video? Extension models moved from Modify Video into Extend Video. Older drafts and canvas cards that used an extension model now open in Extend Video. ## Related - [Extend a shot](https://docs.nodespell.com/recipes/extend-a-shot/) — a worked example - [Modify Video](https://docs.nodespell.com/creative-tools/modify-video/) — change content inside a clip - [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) — start a new shot from a still - [Timeline](https://docs.nodespell.com/edit/timeline/) — trim sources and assemble extended shots - [Supported media](https://docs.nodespell.com/reference/supported-media/) — file types you can upload --- # Raise video resolution with Video Upscaler > Use the Nodespell Video Upscaler Creative Tool to enlarge and restore an existing clip, or render a Seedance 2.5 draft at full quality. Source: https://docs.nodespell.com/creative-tools/video-upscaler/ · Last updated: 2026-11-10 Video Upscaler is the Nodespell Creative Tool for making an existing video larger and cleaner. You supply one clip, choose an upscaling model and its scale or target resolution, and get back a higher-resolution version with the same motion and content. Video Upscaler improves footage; it does not invent new scenes. ## When to use Video Upscaler | You want to… | Use | | ------------------------------------------------------------------------ | --------------------------------------------------- | | Deliver a clip at a higher resolution, or clean up noise and compression | **Video Upscaler** | | Render a Seedance 2.5 **Draft Mode** preview at 1080p | **Video Upscaler** with Seedance 2.5 Draft to Final | | Change what happens in the clip | [Modify Video](https://docs.nodespell.com/creative-tools/modify-video/) | | Enlarge a still image | [Image Upscaler](https://docs.nodespell.com/creative-tools/image-upscaler/) | Upscale last, once the edit is locked. Upscaled files are larger, and many models charge by output resolution, frame rate and length. ## Upscale a video 1. Switch the workspace to **Create**. Select **Image to Video** in the tool rail, open the tool switcher at the top of the form, and choose **Video Upscaler**. It is also under **All tools**. 2. Check the model. The default is **Topaz Video Upscaler**. 3. Add the clip to the video field. Select it to open **Choose assets** and pick an Asset or upload a file. 4. Set the scale factor or target resolution, and any restoration options the model offers. 5. Set **Generations** if you want more than one attempt, from 1 to 4. 6. Select **Upscale**. ## Video Upscaler controls | Control | What it does | Options and defaults | | ----------------------------------- | -------------------------------------------------------- | ----------------------------- | | Tool switcher (title) | Moves to another video Creative Tool | All video tools | | Model picker | Chooses the upscaling model | Default: Topaz Video Upscaler | | **Previous model** / **Next model** | Steps through eligible models | — | | Source video field | The clip to upscale; always required | One video | | **More settings** | Less-used model settings, with a count of changed values | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Upscale** | Adds or updates the model node and submits the runs | Shows the credit estimate | ## Recommended models for Video Upscaler The **Featured** collection lists Topaz Video Upscaler (default) and SeedVR Upscale Video. **All models** adds ByteDance Video Upscaler, Crystal Video Upscaler and Seedance 2.5 Draft to Final. | Model | Main controls | | --------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Topaz Video Upscaler | **Enhancement Model** (default Proteus), **Upscale Factor** 1–4 (default 2), **Target FPS** up to 60 (-1 keeps the source rate), restoration sliders such as **Noise Reduction** and **Recover Detail**, **H.264 Output** | | SeedVR Upscale Video | **Upscale Mode** Target or Factor; **Target Resolution** 720p to 2160p; **Upscale Factor**; output format and quality | | ByteDance Video Upscaler | **Target Resolution** 1080p, 2K or 4K; **Target FPS** 30 or 60; **Enhancement Tier**; **Enhancement Preset** such as UGC or Old Film; **Fidelity** | | Crystal Video Upscaler | **Scale Factor**; keep the output below 5K | | Seedance 2.5 Draft to Final | **Draft Video** and **Output Format** (MP4 or MOV) only | Topaz restoration sliders default to -1, which means "use the selected enhancement model's default". See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) for switching. ## Render a Seedance 2.5 draft at 1080p Seedance 2.5 video models have a **Draft Mode** toggle that produces a 480p preview. To finish a draft you like: 1. Open **Video Upscaler** and choose **Seedance 2.5 Draft to Final** in the model picker. 2. Add the original draft video to **Draft Video**. 3. Choose MP4 or MOV, then select **Upscale**. The final reuses the draft's prompt, references, duration, framing and audio settings. It works only with an original Seedance 2.5 draft generated in Nodespell within the past seven days; uploaded copies and edited videos are not supported. Rendering the final uses additional credits. ## What changes with the model? - **Scale:** factor-based models multiply the source size; target-based models aim for a named resolution. - **Frame rate:** Topaz Video Upscaler and ByteDance Video Upscaler can change the output frame rate up to 60 FPS. Higher frame rates raise the cost. - **Restoration strength:** Topaz offers per-artifact sliders; ByteDance offers tiers and presets; SeedVR offers a noise scale. - **Format:** some models let you choose the container or codec, for example **H.264 Output** on Topaz. ## Where does the upscaled video go? **Upscale** places the model node in the current graph level and runs it. The result appears in **Results** in the Create form and on the node; see [Takes](https://docs.nodespell.com/graph/takes/). The source is unchanged. Generated media also appears under **Generated** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). To download it, see [Export](https://docs.nodespell.com/edit/export/). ## Common problems - **"This value is required."** No source video is set. - **The estimate is much higher than expected.** Output resolution, frame rate, enhancement tier and clip length all raise cost. Lower the factor or frame rate, or trim the clip first. - **Seedance 2.5 Draft to Final fails.** The input is not an original Seedance 2.5 draft from Nodespell, or it is older than seven days. ## Frequently asked questions ### What is the default Video Upscaler model? The default is Topaz Video Upscaler. It upscales by a factor of 1 to 4, default 2, with a choice of enhancement models and optional frame-rate conversion up to 60 FPS. ### Does Video Upscaler change the content of my clip? No. The upscaling models enlarge and restore existing footage while keeping its motion and structure. To change content, use [Modify Video](https://docs.nodespell.com/creative-tools/modify-video/). ### Can I upscale to 4K? Yes, with a model that targets it: SeedVR Upscale Video offers up to 2160p and ByteDance Video Upscaler offers 4K. With factor-based models, the result depends on the source size. ### What is Seedance 2.5 Draft to Final? It renders a 480p Seedance 2.5 **Draft Mode** video at 1080p, reusing the draft's settings. It accepts only original drafts generated in Nodespell within seven days. ## Related - [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) — includes Seedance 2.5 **Draft Mode** - [Image Upscaler](https://docs.nodespell.com/creative-tools/image-upscaler/) — the still-image equivalent - [Export](https://docs.nodespell.com/edit/export/) — get finished video out of Nodespell - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare upscaling models - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) — how estimates work --- # Make a talking video with Speak > Use the Nodespell Speak Creative Tool to make a portrait talk from a script or audio, or lip-sync an existing video to new speech. Source: https://docs.nodespell.com/creative-tools/speak/ · Last updated: 2026-11-10 Speak is the Nodespell Creative Tool for talking videos. It covers two jobs: making a still portrait or character speak from a script or an audio file (avatar models), and re-syncing the mouth of a person in an existing video to new audio (lip-sync models). The selected model decides which inputs you need. ## When to use Speak | You have… | Choose a Speak model that takes… | | ------------------------------------------------- | -------------------------------------------------------- | | A portrait image and a script | An image and a script, such as P-Video Avatar | | A portrait image and recorded or generated speech | An image and audio, such as P-Video Avatar or Omni Human | | A video of someone talking and new audio | A video and audio, such as Sync Lipsync V3 | If you only need the voice, without a picture, use [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/). For a general shot from a still, use [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/). ## Make a portrait speak 1. Switch the workspace to **Create**. Select **Image to Video** in the tool rail, open the tool switcher at the top of the form, and choose **Speak**. It is also under **All tools**. 2. Keep the default model, **P-Video Avatar**, or choose another. 3. Add the portrait to **Source image**. 4. Either type the words in **Voice script**, or add an audio file to **Reference audio**. If you add audio, P-Video Avatar uses it instead of the generated voice. 5. For a script, choose a **Voice** and adjust the voice language and voice prompt if needed. 6. Set **Resolution** (720p or 1080p) and **Generations** from 1 to 4. 7. Select **Generate**. ## Lip-sync an existing video 1. Open **Speak** and choose a lip-sync model, such as **Sync Lipsync V3**. 2. Add the video of the person talking and the new audio track. Both are required on Sync Lipsync V3. 3. Choose how to handle a length mismatch with **Sync Mode**: Cut Off, Loop, Bounce, Silence or Remap. 4. Select **Generate**. To create the new audio first, use [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/) and pick its result in the audio field. See [Make a talking video](https://docs.nodespell.com/recipes/talking-video/) for the full chain. ## Speak controls | Control | What it does | Options and defaults | | ----------------------------------- | --------------------------------------------------- | ------------------------- | | Tool switcher (title) | Moves to another video Creative Tool | All video tools | | Model picker | Chooses an avatar or lip-sync model | Default: P-Video Avatar | | **Previous model** / **Next model** | Steps through eligible models | — | | **More settings** | Less-used model settings | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Generate** | Adds or updates the model node and submits the runs | Shows the credit estimate | ## P-Video Avatar settings | Setting | What it does | Default | | ------------------- | ------------------------------------------------------------- | ------------------------ | | **Source image** | The portrait or character that speaks | Required | | **Reference audio** | Speech to use instead of a generated voice | Optional | | **Voice script** | Words to say when no audio is provided | Empty | | **Resolution** | 720p or 1080p | 720p | | **Voice** | Generated voice preset | Zephyr (Female) | | **Voice Language** | Language or accent, such as English (US), Spanish or Japanese | English (US) | | **Voice Prompt** | How to deliver the script; not spoken | "Say the following." | | **Video Prompt** | How the person should look or behave while speaking | "The person is talking." | ## Recommended models for Speak The **Featured** collection lists P-Video Avatar (default), HeyGen Avatar V and Sync Lipsync V3. **All models** also includes Davinci Magihuman, Omni Human, Kling Lip Sync, Sync Lipsync 2, Sync Lipsync 2 Pro, HeyGen V3 Lipsync Precision and HeyGen V3 Lipsync Speed. | Model | Inputs | | ----------------- | ----------------------------------------------------------------------- | | P-Video Avatar | Image, plus script or audio | | HeyGen Avatar V | Script, plus an existing avatar ID and voice ID from that service | | Sync Lipsync V3 | Video and audio | | Omni Human | Image and audio; its description recommends audio of 15 seconds or less | | Davinci Magihuman | Image and prompt, with optional driving audio; 1–30 seconds | See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) for filters and switching. ## What changes with the model? - **Image or video source:** avatar models animate a still; lip-sync models change an existing clip. - **Script or audio:** some models generate speech from text, others need an audio file. - **Length:** usually follows the audio or script. Some models note limits, such as Kling Lip Sync's 2–10 second source videos. - **Resolution:** set on some models (P-Video Avatar, HeyGen Avatar V); others follow the source. ## Where does the video go? **Generate** places the Speak model node in the current graph level and runs it. The talking video appears in **Results** in the Create form and on the node; see [Takes](https://docs.nodespell.com/graph/takes/). Generated media also appears under **Generated** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ## Common problems - **The mouth does not match the words.** Use a clear, front-facing face with good lighting. Side views and covered mouths reduce quality. - **"This value is required."** A required image, video or audio is missing for the selected model. - **The voice sounds wrong.** Change **Voice**, **Voice Language** or **Voice Prompt**, or record or generate the audio separately and use it as **Reference audio**. ## Frequently asked questions ### Can Speak make a photo talk? Yes. Use an avatar model such as P-Video Avatar, add the photo as the source image, and supply a script or an audio file. ### Can Speak change what someone says in an existing video? Yes, with a lip-sync model such as Sync Lipsync V3. Add the video and the new audio, and the model re-times the mouth to match. ### Do I need to record audio for Speak? No. P-Video Avatar can generate speech from **Voice script**. Lip-sync models need an audio file, which you can make in [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/). ### What is the default Speak model? The default is P-Video Avatar, which turns a portrait image plus a script or audio into a talking video at 720p or 1080p. ## Related - [Make a talking video](https://docs.nodespell.com/recipes/talking-video/) — script to voice to speaking video - [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/) — create the speech track - [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) — animate a still without speech - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare avatar and lip-sync models - [Timeline](https://docs.nodespell.com/edit/timeline/) — place the talking clip in an edit --- # Create speech, music or sound with Audio Generator > Use the Nodespell Audio Generator Creative Tool to make speech, music or sound effects from a prompt, with every audio model in one picker. Source: https://docs.nodespell.com/creative-tools/audio-generator/ · Last updated: 2026-11-10 Audio Generator is the Nodespell Creative Tool that brings every audio-making model into one place. It lists speech, music and sound-effect models together, so you can make a voiceover, a song or a sound effect from a prompt without first deciding which focused tool to open. The model you pick decides what kind of audio you get. ## Audio Generator or a focused audio tool? Audio Generator is the umbrella. The three focused tools each show only one kind of model, which makes the picker shorter. | You want… | Use | | ------------------------------------------------------------------- | ------------------------------------------------------------------- | | To browse all audio models, or you are not sure which kind you need | **Audio Generator** | | Spoken words from a script | [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/) | | A song, score or instrumental | [Music Generator](https://docs.nodespell.com/creative-tools/music-generator/) | | A sound effect or ambience | [Sound Effects Generator](https://docs.nodespell.com/creative-tools/sound-effects-generator/) | | A face that speaks | [Speak](https://docs.nodespell.com/creative-tools/speak/) (makes video, not audio) | The same model gives the same result in Audio Generator and in its focused tool. Only the list of models differs. ## Generate audio 1. Switch the workspace to **Create** and select **Audio Generator** in the tool rail. Audio Generator has its own rail button. 2. Check the model. The default is **ElevenLabs TTS V3**, a text-to-speech model. 3. To make music or sound effects instead, open the model picker and choose a music or sound-effect model. 4. Fill in the prompt or text field the model shows, plus any voice, style or format settings. 5. Set **Generations** from 1 to 4. 6. Select **Generate**. You can also switch to Voice Generator, Music Generator or Sound Effects Generator from the tool switcher at the top of the form. ## Audio Generator controls | Control | What it does | Options and defaults | | ----------------------------------- | -------------------------------------------------------- | -------------------------------------------------------------------------- | | Tool switcher (title) | Moves to another audio Creative Tool | Audio Generator, Voice Generator, Music Generator, Sound Effects Generator | | Model picker | Chooses any eligible speech, music or sound-effect model | Default: ElevenLabs TTS V3 | | **Previous model** / **Next model** | Steps through all eligible audio models | — | | **More settings** | Less-used model settings | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Generate** | Adds or updates the model node and submits the runs | Shows the credit estimate | ## Recommended models for Audio Generator The **Featured** collection lists, in order: | Model | Kind | | --------------------------- | ------------------------------------------- | | ElevenLabs TTS V3 (default) | Speech | | Gemini 3.1 Flash TTS | Speech, including multi-speaker scripts | | Google Lyria 3 Pro | Music, full songs | | Minimax Music 2.6 | Music, with optional lyrics or instrumental | | Google Lyria 3 | Music, 30-second clips | **All models** also lists sound-effect models such as ElevenLabs Sound Effects V2 and MMAudio V2, and further speech and music models. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). ## What changes with the model? Because Audio Generator mixes three kinds of model, the form changes more than in any other tool when you switch. - **Speech models** show a text field and voice settings, such as **Voice**, **Speed** and **Stability** on ElevenLabs TTS V3. - **Music models** show a description prompt, and some add lyrics, an instrumental switch, or images for inspiration. - **Sound-effect models** show a description and, on some models, a duration or loop option. - **Output format** varies: some models let you choose MP3, WAV or other formats, others return one fixed format. The focused tool pages describe each kind in detail: [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/), [Music Generator](https://docs.nodespell.com/creative-tools/music-generator/) and [Sound Effects Generator](https://docs.nodespell.com/creative-tools/sound-effects-generator/). ## Where does the audio go? **Generate** places the model node in the current graph level and runs it. Finished audio appears in **Results** in the Create form, where you can play it and step through older and newer results; the node keeps the full history, described in [Takes](https://docs.nodespell.com/graph/takes/). Generated audio also appears under **Generated** and **Audio** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). To place it under a picture, use the [Timeline](https://docs.nodespell.com/edit/timeline/). ## Common problems - **The result is speech when you wanted music.** The default model is a speech model. Choose a music model in the picker, or open Music Generator. - **"This value is required."** The model's prompt or text field is empty. - **The prompt field says "Prompt controlled by an input node".** A connected node supplies the text. Edit that node or select **Disconnect**. ## Frequently asked questions ### What is the difference between Audio Generator and Voice Generator? Audio Generator lists every speech, music and sound-effect model. Voice Generator lists only speech models, so its picker is shorter; the models behave the same in both. ### What is the default Audio Generator model? The default is ElevenLabs TTS V3, a text-to-speech model. Choose a music or sound-effect model in the picker for other kinds of audio. ### Can Audio Generator make audio from a video? Some models can. MMAudio V2 accepts an optional video for video-to-audio generation, and Sonilo V1.1 Video to Music makes music from a video. ### How do I add the audio to a video? Place both in the [Timeline](https://docs.nodespell.com/edit/timeline/). The [soundtrack recipe](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/) walks through it. ## Related - [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/) — speech only - [Music Generator](https://docs.nodespell.com/creative-tools/music-generator/) — music only - [Sound Effects Generator](https://docs.nodespell.com/creative-tools/sound-effects-generator/) — effects only - [Add a soundtrack to a clip](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/) — music and effects under video - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — picker and switching - [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/) — the same node in Graph --- # Turn a script into speech with Voice Generator > Use the Nodespell Voice Generator Creative Tool to turn a written script into spoken audio, choosing a voice, language and delivery style. Source: https://docs.nodespell.com/creative-tools/voice-generator/ · Last updated: 2026-11-10 Voice Generator is the Nodespell Creative Tool for text-to-speech. You write a script, pick a speech model and a voice, and get back an audio file of the script being spoken. Voice Generator lists only speech models; for music or sound effects, use the other audio tools. ## When to use Voice Generator | You want… | Use | | --------------------------------------------------- | ------------------------------------------------------------------------- | | A voiceover, narration or line of dialogue as audio | **Voice Generator** | | To browse speech, music and effects models together | [Audio Generator](https://docs.nodespell.com/creative-tools/audio-generator/) | | A face that says the line | [Speak](https://docs.nodespell.com/creative-tools/speak/), optionally using Voice Generator's audio | | A song with sung lyrics | [Music Generator](https://docs.nodespell.com/creative-tools/music-generator/) | ## Generate speech 1. Switch the workspace to **Create**. Select **Audio Generator** in the tool rail, open the tool switcher at the top of the form, and choose **Voice Generator**. It is also under **All tools**. 2. Check the model. The default is **ElevenLabs TTS V3**. 3. Type the script in the text field. 4. Choose a **Voice**, and adjust delivery settings such as speed or style if the model offers them. 5. Set **Generations** from 1 to 4 if you want several reads to choose from. 6. Select **Generate**. ## Voice Generator controls | Control | What it does | Options and defaults | | ----------------------------------- | --------------------------------------------------- | -------------------------------------------------------------------------- | | Tool switcher (title) | Moves to another audio Creative Tool | Audio Generator, Voice Generator, Music Generator, Sound Effects Generator | | Model picker | Chooses a speech model | Default: ElevenLabs TTS V3 | | **Previous model** / **Next model** | Steps through speech models | — | | **More settings** | Less-used model settings | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Generate** | Adds or updates the model node and submits the runs | Shows the credit estimate | ## Recommended models for Voice Generator The **Featured** collection lists ElevenLabs TTS V3 (default) and Gemini 3.1 Flash TTS. **All models** also includes ElevenLabs TTS Turbo V2.5, Kling Text To Speech, Kokoro 8.2m, Minimax Speech 02 HD, Minimax Speech 02 Turbo, Qwen3 TTS, Resemble AI Chatterbox and Resemble AI Chatterbox Pro. ### ElevenLabs TTS V3 settings | Setting | What it does | Default | | -------------------------------- | -------------------------------------------------------------------- | --------------- | | **Text** | The script to speak | Empty | | **Voice** | Preset voice, listed by gender and accent, such as "F - US - Rachel" | F - US - Rachel | | **Speed** | 0.7 to 1.2; below 1 is slower | 1 | | **Stability** | 0 to 1; lower is more expressive, higher is more consistent | 0.5 | | **Similarity Boost** | 0 to 1; how closely to follow the original voice | 0.75 | | **Style** | 0 to 1; exaggerates the voice's speaking style | 0 | | **Previous Text**, **Next Text** | Surrounding lines that help continuity when you split a long script | Empty | ### Gemini 3.1 Flash TTS settings Gemini 3.1 Flash TTS reads inline delivery tags in the prompt, such as `[laughing]`, `[whispering]` or `[short pause]`. **Style Instructions** describe tone, pace and accent without being spoken. For dialogue, add **Speakers** entries whose **Speaker Alias** matches the line prefix in the script, for example `Host:`, and give each its own **Voice**. **Language Code** defaults to English (US). ## What changes with the model? - **Voices:** every model has its own voice list; voices do not carry over when you switch. - **Languages:** some models offer a language setting; check the model's description for supported languages. - **Reference audio:** a few models, such as Qwen3 TTS and Resemble AI Chatterbox, accept an audio sample to guide the voice. - **Multiple speakers:** Gemini 3.1 Flash TTS supports multi-speaker scripts; most models read one voice. When you switch models, Nodespell carries over the settings it can map and tells you which could not be carried over. ## Where does the audio go? **Generate** places the speech model node in the current graph level and runs it. The audio appears in **Results** in the Create form and stays in the node's history; see [Takes](https://docs.nodespell.com/graph/takes/). Generated audio also appears under **Generated** and **Audio** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/), ready to use in [Speak](https://docs.nodespell.com/creative-tools/speak/) or the [Timeline](https://docs.nodespell.com/edit/timeline/). ## Common problems - **Words are mispronounced.** Spell them phonetically in the script, or try another voice or model. - **The read sounds flat or erratic.** On ElevenLabs TTS V3, lower **Stability** for more emotion or raise it for a steadier read. - **Long scripts drift between parts.** Split the script and use **Previous Text** and **Next Text** so each part knows its neighbors. ## Frequently asked questions ### What is the default Voice Generator model? The default is ElevenLabs TTS V3. It has a long list of preset voices and controls for speed, stability, similarity and style. ### Can Voice Generator create a dialogue with two voices? Yes, with Gemini 3.1 Flash TTS. Add a speaker entry for each alias used in the script and choose a voice for each. ### Can I clone my own voice? Some models accept reference audio to guide the voice, such as Qwen3 TTS and Resemble AI Chatterbox. Use audio you have the right to use. ### How do I make a character say the line on screen? Generate the audio here, then use it as the audio input in [Speak](https://docs.nodespell.com/creative-tools/speak/). ## Related - [Audio Generator](https://docs.nodespell.com/creative-tools/audio-generator/) — all audio models in one picker - [Speak](https://docs.nodespell.com/creative-tools/speak/) — put the voice on a face - [Make a talking video](https://docs.nodespell.com/recipes/talking-video/) — script to voice to video - [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/) — reuse scripts - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare speech models --- # Compose music with Music Generator > Use the Nodespell Music Generator Creative Tool to create songs, scores or instrumentals from a description, with optional lyrics and references. Source: https://docs.nodespell.com/creative-tools/music-generator/ · Last updated: 2026-11-10 Music Generator is the Nodespell Creative Tool for making music from a description. You describe the genre, mood, instruments and structure, optionally add lyrics or reference media, and the selected music model returns an audio track. Music Generator lists only music models. ## When to use Music Generator | You want… | Use | | ------------------------------------- | ------------------------------------------------------------------- | | A song, instrumental bed or score cue | **Music Generator** | | Spoken narration | [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/) | | A sound effect or ambience | [Sound Effects Generator](https://docs.nodespell.com/creative-tools/sound-effects-generator/) | | To browse every audio model at once | [Audio Generator](https://docs.nodespell.com/creative-tools/audio-generator/) | ## Generate music 1. Switch the workspace to **Create**. Select **Audio Generator** in the tool rail, open the tool switcher at the top of the form, and choose **Music Generator**. It is also under **All tools**. 2. Check the model. The default is **Google Lyria 3 Pro**. 3. Write the prompt: genre, mood, tempo, instruments, and the structure you want. 4. Add lyrics, an instrumental setting or reference images if the model offers them. 5. Set **Generations** from 1 to 4 to get several takes on the idea. 6. Select **Generate**. ## Music Generator controls | Control | What it does | Options and defaults | | ----------------------------------- | --------------------------------------------------- | -------------------------------------------------------------------------- | | Tool switcher (title) | Moves to another audio Creative Tool | Audio Generator, Voice Generator, Music Generator, Sound Effects Generator | | Model picker | Chooses a music model | Default: Google Lyria 3 Pro | | **Previous model** / **Next model** | Steps through music models | — | | **More settings** | Less-used model settings | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Generate** | Adds or updates the model node and submits the runs | Shows the credit estimate | ## Recommended models for Music Generator The **Featured** collection lists, in order: | Model | Good for | Main inputs | | ---------------------------- | ---------------------------------------------- | --------------------------------------------------------------------------------------- | | Google Lyria 3 Pro (default) | Full-length songs, up to around 3 minutes | Prompt (required); up to 10 images for inspiration | | Minimax Music 2.6 | Vocal songs with your lyrics, or instrumentals | Prompt, **Lyrics**, **Instrumental**, **Auto Lyrics**, **Audio Format** (MP3, WAV, PCM) | | Google Lyria 3 | 30-second music clips | Prompt (required); up to 10 images for inspiration | **All models** also lists Google Lyria 2, Minimax Music 01, 1.5 and 2.5, Sonilo V1.1 Text to Music, and Sonilo V1.1 Video to Music, which scores an input video. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). ## Shape the structure with tags Google Lyria 3 Pro and Google Lyria 3 read section tags in the prompt, such as `[Intro]`, `[Verse]`, `[Chorus]`, `[Bridge]` and `[Outro]`, and timeline tags such as `[0:00 - 0:30]`. Minimax Music 2.6 reads similar tags inside **Lyrics**, one line per lyric line. Tags guide the arrangement; they do not guarantee exact timing. ## What changes with the model? - **Length:** Google Lyria 3 makes 30-second clips; Google Lyria 3 Pro makes songs of up to around 3 minutes; Sonilo V1.1 Text to Music has a duration setting. - **Vocals and lyrics:** Minimax Music 2.6 can sing your lyrics, write its own with **Auto Lyrics**, or make an instrumental. On that model, **Instrumental** requires a prompt. - **References:** the Lyria 3 models take images for inspiration; Sonilo V1.1 Video to Music takes a video; Minimax Music 01 takes audio files. - **Format:** some models let you choose format, bitrate and sample rate; others return one format. ## Where does the music go? **Generate** places the music model node in the current graph level and runs it. The track appears in **Results** in the Create form and in the node's history; see [Takes](https://docs.nodespell.com/graph/takes/). Generated audio also appears under **Generated** and **Audio** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). Place it under a picture in the [Timeline](https://docs.nodespell.com/edit/timeline/). ## Common problems - **"This value is required."** The prompt is empty on a model that requires one. - **The song structure ignores your sections.** Put each section tag on its own line, and keep the description and tags consistent. - **You hear vocals in what should be a bed.** Use a model with an instrumental option, such as Minimax Music 2.6, or say "instrumental, no vocals" in the prompt. ## Frequently asked questions ### What is the default Music Generator model? The default is Google Lyria 3 Pro. It generates full-length songs of up to around 3 minutes from a prompt and up to 10 inspiration images. ### Can Music Generator sing my lyrics? Yes, with Minimax Music 2.6 or another model with a lyrics field. Add structure tags such as `[Verse]` and `[Chorus]` to guide the arrangement. ### Can I make music to fit a video? Sonilo V1.1 Video to Music takes an input video and generates music for it. Otherwise, describe the mood and length in the prompt and align the track in the [Timeline](https://docs.nodespell.com/edit/timeline/). ### How long can a generated track be? It depends on the model: about 30 seconds on Google Lyria 3, up to around 3 minutes on Google Lyria 3 Pro. ## Related - [Add a soundtrack to a clip](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/) — music and effects under video - [Audio Generator](https://docs.nodespell.com/creative-tools/audio-generator/) — all audio models together - [Sound Effects Generator](https://docs.nodespell.com/creative-tools/sound-effects-generator/) — effects and ambience - [Timeline](https://docs.nodespell.com/edit/timeline/) — place and trim the track - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare music models --- # Create sound effects with Sound Effects Generator > Use the Nodespell Sound Effects Generator Creative Tool to create sound effects and ambience from a text description or from a video clip. Source: https://docs.nodespell.com/creative-tools/sound-effects-generator/ · Last updated: 2026-11-10 Sound Effects Generator is the Nodespell Creative Tool for making sound effects and ambience. You describe the sound—footsteps on gravel, a door creaking, a busy street—and the selected model returns an audio clip. One model can also listen to a video and generate sound that fits the picture. ## When to use Sound Effects Generator | You want… | Use | | ----------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- | | A sound effect, foley or ambience | **Sound Effects Generator** | | Sound that matches the action in a clip | **Sound Effects Generator** with MMAudio V2 and a video | | Music | [Music Generator](https://docs.nodespell.com/creative-tools/music-generator/) | | Speech | [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/) | | Synchronized audio while generating a new video | A video model with a **Generate Audio** setting, in [Text to Video](https://docs.nodespell.com/creative-tools/text-to-video/) or [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) | ## Generate a sound effect 1. Switch the workspace to **Create**. Select **Audio Generator** in the tool rail, open the tool switcher at the top of the form, and choose **Sound Effects Generator**. It is also under **All tools**. 2. Check the model. The default is **ElevenLabs Sound Effects V2**. 3. Describe the sound in the text field. Be concrete about the source, the surface and the space, for example "heavy boots on wet gravel, outdoors, light rain". 4. Set duration, looping or prompt influence if needed. 5. Set **Generations** from 1 to 4 to get variations to choose from. 6. Select **Generate**. ## Sound Effects Generator controls | Control | What it does | Options and defaults | | ----------------------------------- | --------------------------------------------------- | -------------------------------------------------------------------------- | | Tool switcher (title) | Moves to another audio Creative Tool | Audio Generator, Voice Generator, Music Generator, Sound Effects Generator | | Model picker | Chooses a sound-effect model | Default: ElevenLabs Sound Effects V2 | | **Previous model** / **Next model** | Switches between the two models | — | | **More settings** | Less-used model settings | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Generate** | Adds or updates the model node and submits the runs | Shows the credit estimate | ## Models for Sound Effects Generator Sound Effects Generator has no featured list, so the picker opens on **All models**. It lists two models. ### ElevenLabs Sound Effects V2 (default) | Setting | What it does | Default | | -------------------- | ------------------------------------------------------------------ | ------------------- | | **Text** | Description of the sound | Empty | | **Duration Seconds** | 0.5 to 22 seconds; -1 lets the model choose from the prompt | -1 | | **Loop** | Makes a sound that loops smoothly | Off | | **Prompt Influence** | 0 to 1; higher follows the prompt more closely with less variation | 0.3 | | **Output Format** | MP3, PCM or Opus at several sample rates and bitrates | MP3 44.1kHz 128kbps | ### MMAudio V2 MMAudio V2 generates audio from an optional video, an optional image, or a text prompt. Add a video to get sound timed to the action on screen. Its settings include **Duration** (default 8 seconds), **Negative Prompt** (default "music", to keep music out), **Num Steps**, **CFG Strength** and **Seed**. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) for switching. ## Add sound to a video clip 1. Open **Sound Effects Generator** and choose **MMAudio V2**. 2. Add the clip to the video field. Select it to open **Choose assets**. 3. Optionally describe the sound you want, and set **Duration** to match the clip. 4. Select **Generate**, then place the audio under the clip in the [Timeline](https://docs.nodespell.com/edit/timeline/). ## Where does the sound go? **Generate** places the model node in the current graph level and runs it. The sound appears in **Results** in the Create form and in the node's history; see [Takes](https://docs.nodespell.com/graph/takes/). Generated audio also appears under **Generated** and **Audio** in the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ## Common problems - **The sound is the wrong length.** On ElevenLabs Sound Effects V2, set **Duration Seconds** instead of -1. Its maximum is 22 seconds. - **Music creeps into the effect.** Keep **Negative Prompt** set to "music" on MMAudio V2, or remove musical words from the description. - **A loop clicks at the join.** Turn on **Loop** on ElevenLabs Sound Effects V2 so the model builds a smooth loop. ## Frequently asked questions ### What is the default Sound Effects Generator model? The default is ElevenLabs Sound Effects V2. It makes effects from a text description, from 0.5 to 22 seconds long, with an optional smooth loop. ### Can Sound Effects Generator create sound from a video? Yes. Choose MMAudio V2 and add a video; it generates audio that follows the picture. ### Can I make a longer ambience bed? ElevenLabs Sound Effects V2 stops at 22 seconds. Turn on **Loop** and repeat the clip in the [Timeline](https://docs.nodespell.com/edit/timeline/) to cover a longer shot. ### Is Sound Effects Generator the same as Audio Generator? No. Audio Generator lists speech, music and sound-effect models together. Sound Effects Generator lists only sound-effect models. ## Related - [Add a soundtrack to a clip](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/) — music and effects under video - [Audio Generator](https://docs.nodespell.com/creative-tools/audio-generator/) — all audio models together - [Music Generator](https://docs.nodespell.com/creative-tools/music-generator/) — music beds and songs - [Timeline](https://docs.nodespell.com/edit/timeline/) — place, trim and layer sounds - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — the model picker --- # Write or transform text with Generate Text > Use the Nodespell Generate Text Creative Tool to write, rewrite or describe text with a language model, from a prompt and optional images. Source: https://docs.nodespell.com/creative-tools/generate-text/ · Last updated: 2026-11-10 Generate Text is the Nodespell Creative Tool for creating or transforming text with a language model. You write a prompt—and can add images for the model to look at—and get back text: a script, a set of prompts, a caption, a description of a picture. The result is a text output you can read, copy or connect to other nodes. ## When to use Generate Text | You want… | Use | | ----------------------------------------------------------------- | --------------------------------------------------- | | A script, shot list, caption or rewritten prompt | **Generate Text** | | A description of an image | **Generate Text** with the image added | | Help building or changing the graph, or advice about your project | [Agent](https://docs.nodespell.com/agent/) | | Spoken audio of a script | [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/) | Generate Text runs one model on one prompt and returns its answer. It does not read your graph or make changes to it; the Agent does that. ## Generate text 1. Switch the workspace to **Create** and open **All tools**, then select **Generate Text**. Generate Text has no button of its own in the tool rail. 2. Check the model. The default is **GPT Text**. 3. Write the prompt. To have the model look at pictures, add them to the images field. 4. Optionally change **System Instructions** to give the model a role or rules, such as "You write short, visual video prompts." 5. Set **Generations** from 1 to 4 if you want several versions. 6. Select **Generate**. ## Generate Text controls | Control | What it does | Options and defaults | | ----------------------------------- | -------------------------------------------------------- | ------------------------- | | Model picker | Chooses the text model | Default: GPT Text | | **Previous model** / **Next model** | Steps through text models | — | | **More settings** | Less-used model settings, with a count of changed values | Collapsed | | **Generations** | Number of runs submitted together | 1 to 4, default 1 | | **Generate** | Adds or updates the model node and submits the runs | Shows the credit estimate | ## Recommended models for Generate Text The **Featured** collection lists GPT Text (default), Claude Text and Gemini Text. **All models** also lists Deepseek R1 and Deepseek V3. Each featured model has a **Model** setting that picks a version within that family: | Model | **Model** options | Default | | ----------- | ------------------------------------------------------- | --------------------- | | GPT Text | GPT 5 Nano, GPT 5 Mini, GPT 5 | GPT 5 Nano | | Claude Text | Claude 3.5 Haiku, Claude 4.5 Sonnet, Claude 4.1 Opus | Claude 3.5 Haiku | | Gemini Text | Gemini 2.5 Flash Lite, Gemini 2.5 Flash, Gemini 2.5 Pro | Gemini 2.5 Flash Lite | All three featured models share these settings: | Setting | What it does | Default | | ----------------------- | ------------------------------------------------- | ------------------------------ | | **Prompt** | The request or text to process | — | | **System Instructions** | The persona or rules the model follows | "You are a helpful assistant." | | Maximum output tokens | Caps the length of the answer, up to 3,000 tokens | 1000 | | Images input | Pictures the model can read | Optional | The output-length setting is labelled **Max Output Tokens** on GPT Text and Gemini Text, and **Maximum Output Tokens** on Claude Text. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) for switching. ## Use the text in your graph Generate Text places a text model node in the current graph level. In Graph you can connect its text output to the prompt input of an image, video or audio node, so the generated text drives the next step. When a prompt is fed this way, the Create form shows "Prompt controlled by an input node" on the receiving tool. See [Connections](https://docs.nodespell.com/graph/connections/) and [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/). ## Where does the text go? The answer appears in **Results** in the Create form and stays in the node's history; see [Takes](https://docs.nodespell.com/graph/takes/). To read a long answer, open it in the Design panel, which shows text with search, wrapping and Markdown views; see [Viewers](https://docs.nodespell.com/edit/viewers/). ## Common problems - **The answer stops mid-sentence.** Raise the output-token setting, up to 3,000. - **The answer ignores the image.** Make sure the image is in the images field, and ask about it directly in the prompt. - **Answers are too generic.** Put the format and constraints in **System Instructions**, and give an example in the prompt. ## Frequently asked questions ### What is the default Generate Text model? The default is GPT Text, using GPT 5 Nano unless you change its **Model** setting. Claude Text and Gemini Text are also featured. ### Can Generate Text describe an image? Yes. GPT Text, Claude Text and Gemini Text accept images; add them to the images field and ask about them in the prompt. ### How long can a Generate Text answer be? GPT Text, Claude Text and Gemini Text allow up to 3,000 output tokens, with a default of 1,000. ### Is Generate Text the same as the Agent? No. Generate Text runs one model on your prompt and returns text. The [Agent](https://docs.nodespell.com/agent/) can inspect your project and propose or make graph changes. ### Can I use generated text as a prompt for an image or video? Yes. Connect the Generate Text node's output to the prompt input of another node in Graph. ## Related - [Agent](https://docs.nodespell.com/agent/) — an assistant that works with your graph - [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/) — prompt nodes and reusable text - [Connections](https://docs.nodespell.com/graph/connections/) — feed text into other nodes - [Data and logic nodes](https://docs.nodespell.com/graph/data-and-logic/) — process text further - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — compare text models --- # Navigate and select in the Graph Editor > Move around the Nodespell Graph Editor with a trackpad, mouse or keyboard, select nodes, use the minimap and View Settings, and read node parts. Source: https://docs.nodespell.com/graph/graph-editor/ · Last updated: 2026-11-10 The Graph Editor is the canvas in Nodespell where you place nodes, connect them and run them. Every gesture in the Graph Editor has a trackpad, mouse and keyboard path, so you can pan, zoom, select and frame your graph without a middle mouse button. ## Pan and zoom the canvas Nodespell detects whether you are using a trackpad or a mouse and reads the gesture to match. Turn on **Status Bar Hotkey Help** in **View Settings** to see the gestures for your current device in the status bar. **Trackpad** | To | Do this | | ---------------------- | -------------------------------------- | | Pan | Drag with two fingers | | Zoom | Pinch | | Pan without a gesture | Arrow keys; hold `Shift` to pan faster | | Zoom without a gesture | `=` to zoom in, `-` to zoom out | **Mouse** | To | Do this | | --------------------------- | --------------------------------------------------------------------- | | Zoom | Scroll the wheel | | Pan | Drag with the middle button | | Pan without a middle button | Hold `Cmd` (macOS) or `Ctrl` (Windows) and drag with the right button | | Zoom by dragging | Hold `Shift` and drag with the right button, up/down or left/right | **Natural Scrolling** (Preferences, **Node Graph Editor** section) reverses the scroll direction. Open Preferences with `Cmd/Ctrl+,`. ## Frame nodes and jump around These keys work on any device and act on the current Level. | Key | Action | | --------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------ | | `F` | **Frame Selected Nodes** | | `A` | **Frame All Nodes** | | Hold `Z` | Zoom out to show the whole Level. Release `Z` to zoom in where the pointer is, or drag a box while holding `Z` to frame that region. | | `0` | **Reset Zoom** | | `Shift+0` | **Reset Pan** | | `Shift+[` / `Shift+]` | Previous view / next view | | `Option/Alt+1` … `Option/Alt+0` | Remember the selected nodes in a numbered slot | | `Cmd/Ctrl+Shift+1` … `Cmd/Ctrl+Shift+0` | Jump to the nodes in that slot | You can also right-click empty canvas and choose **Frame Selected Nodes** or **Frame All Nodes**. The full list of keys is on [Keyboard shortcuts](https://docs.nodespell.com/reference/shortcuts/). ## Use the minimap The minimap is a small overview of the nodes, groups and visible area of the current Level. It is off by default. 1. Open **View Settings** in the Graph Editor's menu bar and turn on **Minimap**. 2. Click or drag inside the minimap to move the view there. With the minimap focused, arrow keys pan the graph and `Shift` pans a full screen. 3. Drag the resize handle to make the minimap larger or smaller. The handle also responds to arrow keys. 4. Use the **Previous view** and **Next view** buttons to step back and forward through places you have looked at. 5. Click the pin button to switch between **Auto-hide minimap when not in use** and **Keep minimap open**. To show the minimap in every graph editor you open, turn on **Show Minimap** in Preferences under **Node Graph Editor**. **Auto-hide Minimap** is on by default. ## Select nodes | To | Do this | | --------------------------------------------------- | ------------------------------------ | | Select one node | Click it | | Add or remove a node from the selection | `Shift`+click | | Select an area | Drag on empty canvas (box selection) | | Toggle nodes in an area | `Shift`+drag on empty canvas | | Switch between box and lasso selection | `Shift+B` | | Select everything on the Level | `Cmd/Ctrl+A` | | Clear the selection | `Esc` | | Select everything upstream, downstream or connected | `Shift+U`, `Shift+D`, `Shift+C` | | Step through earlier selections | `[` and `]` | | Walk the selection along connections | `Cmd/Ctrl` + arrow keys | Clicking and dragging is always done with the primary button, so a trackpad click works for every selection action. ## Move, align and duplicate nodes - Drag a node to move it. **Snap To Other Nodes** (on by default) lines its edges or centre up with nearby nodes. - Nudge the selection with `Option/Alt` + arrow keys; add `Shift` for larger steps. - Hold `Option/Alt` while dragging to drag out a copy. - `H` and `V` align the selection horizontally or vertically; `Shift+H` and `Shift+V` distribute it evenly. - `Shift+=` and `Shift+-` make the selected nodes larger or smaller. - `L` tidies the selection or the whole Level. See [Organize your graph](https://docs.nodespell.com/graph/organize/). Undo any edit with `Cmd/Ctrl+Z` and redo with `Cmd/Ctrl+Shift+Z`. ## What are the parts of a node? | Part | What it is | | -------------- | ------------------------------------------------------------------------------------- | | Input sockets | Where connections arrive. In the default top-to-bottom flow they sit on the top edge. | | Output sockets | Where results leave. In the default flow they sit on the bottom edge. | | Body | The node's preview: an image, video, text or audio result, or its prompt text. | | Name tag | The node's name, shown above it when **Node Name Tags** is on. | | Badges | Small markers for state such as errors or warnings, shown when **Node Badges** is on. | | Comment | A note you add with **Edit Comment**, shown when **Node Comment** is on. | Socket colour shows the content type: image, video, audio, text, 3D or any. Hover a socket to see its name, direction and type. When you change the [flow direction](https://docs.nodespell.com/graph/levels-and-folders/), sockets move to the matching edges; node content stays upright and connections keep working. ## What is in View Settings? **View Settings** sits in the Graph Editor's menu bar next to **Nodes**. | Section | Setting | Effect | | ------------ | --------------------------------- | ------------------------------------------------- | | Display | **Minimap** | Shows the overview map | | Display | **Background Grid** | Shows the canvas grid | | Display | **Docked Parameter Editor** (`Q`) | Shows Parameters docked beside the graph | | Node Display | **Node Name Tags** | Shows node names above nodes | | Node Display | **Node Badges** | Shows status badges | | Node Display | **Node Comment** | Shows node comments | | Node Display | **Node Glance** | Shows quick information when you hover nodes | | Interaction | **Reference Links** | Draws links between nodes and what they reference | | Interaction | **Snap To Other Nodes** | Snaps while dragging or resizing | | Interaction | **Keystroke Indicator** | Shows the last shortcut you pressed | | Interaction | **Status Bar Hotkey Help** | Lists keys and gestures for what you are doing | `Shift+S` cycles the connection line style. Grid style, line type and socket colouring are also in Preferences under **Node Graph Editor**. ## Frequently asked questions ### Can I use the Graph Editor without a middle mouse button? Yes. On a trackpad, drag with two fingers to pan and pinch to zoom. With a mouse, hold `Cmd` (macOS) or `Ctrl` (Windows) and right-drag to pan, and use the arrow keys, `F`, `A` and `Z` to move and frame. ### How do I find a node I cannot see? Press `A` to frame every node on the Level, hold `Z` for a quick overview, or open the minimap from **View Settings**. To search by name, use [Find in graph](https://docs.nodespell.com/graph/organize/) with `Cmd/Ctrl+F`. ### Does changing the flow direction move my nodes? No. Changing the flow direction moves sockets to different edges but leaves node positions as they are. Press `L` afterwards if you want Nodespell to rearrange the Level. ### Can I change the keyboard shortcuts? No. Preferences has a **Hotkeys** section where you can view and search every shortcut, but the keys cannot be reassigned. ## Related - [Add nodes](https://docs.nodespell.com/graph/add-nodes/): the Tab menu, Node Browser and Assets - [Connect nodes](https://docs.nodespell.com/graph/connections/): sockets, content types and cutting connections - [Edit parameters](https://docs.nodespell.com/graph/parameters/): the Parameters panel beside the graph - [Organize your graph](https://docs.nodespell.com/graph/organize/): groups, tidying and Find in graph - [Keyboard shortcuts](https://docs.nodespell.com/reference/shortcuts/): every Graph Editor key --- # Add nodes to your graph > Add nodes in Nodespell with the Tab menu, Quick Add, the model hub, the Node Browser or the Asset Manager, preview image effects, and copy or duplicate nodes. Source: https://docs.nodespell.com/graph/add-nodes/ · Last updated: 2026-11-10 The Tab menu is the fastest way to add a node in Nodespell: press `Tab` over the Graph Editor, type what you want, and press `Enter`. Nodespell also adds nodes from the right-click Quick Add menu, the model hub, the Node Browser, and the Asset Manager, and copies or duplicates nodes you already have. ## Add a node with the Tab menu 1. Point at the spot on the canvas where the node should go and press `Tab`. You can also choose **Open Quick Node Menu** from the **Nodes** menu. 2. Type part of a node name, model name or task. The menu searches nodes, models, Creative Tools, custom nodes and workflows at once. 3. Move through results with the arrow keys. The panel on the right previews the highlighted result. 4. Press `Enter` or click a result to create it. Press `Esc` to close without adding anything. If one node is selected when you open the Tab menu, the new node is created next to it and connected to it automatically. If several nodes are selected, hold `Shift` while you choose a node to create and connect one copy for each selected node. ## Browse by category instead of searching With the search field empty, the Tab menu shows category tabs and a row of **Common nodes**. | Tab | Contents | | ------------------- | ----------------------------------------------------------------------- | | **Image models** | Models that output images | | **Video models** | Models that output video | | **Audio models** | Models that output audio | | **Text models** | Models that output text | | **Realtime effect** | Image effects that render in the editor, such as blurs and colour tools | | **Transform** | Nodes that reshape or reposition media | | **Advanced** | Data and logic building blocks | Creative Tools appear in their own **Creative tools** group at the top of a tab. When a node is selected, the empty Tab menu suggests nodes that accept what that node outputs. The footer has filters: **Filter** narrows results by output type (for example **Image models**), **Input** narrows them by what they accept (for example **Accepts image input**), and the **Advanced nodes** switch shows or hides Data, Data Conversion, Flow Control, Logic, Math, Reference and Type Of nodes. See [Data and logic nodes](https://docs.nodespell.com/graph/data-and-logic/) for what those do. Click the star on a result to add it to your favourites. Favourites and recently used nodes appear in the right-click Quick Add menu. ## Preview image effects before adding them Nodespell renders realtime image effects on your own image inside the Tab menu, so you can compare before choosing. 1. Select a node that outputs an image, such as an Asset node holding a photo or an image model with a result. 2. Press `Tab` and open the **Realtime effect** tab, or search for an effect. 3. Highlight an effect. The preview panel shows **Before** and **After** halves while it renders. 4. Move the pointer across the preview to slide the split between the original and the effect. 5. Press `Enter` to add the effect, already connected to the selected node. Previews use your saved defaults for that node type. Effects that need another required input do not show a preview. ## Add a node from the right-click menu Right-click empty canvas to open a menu of shortcuts: | Item | What it does | | -------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Search all nodes** (`Tab`) | Opens the Tab menu | | **Prompt** (`P`) | Creates a Prompt node at the pointer | | **Folder** | Creates a Folder for organizing nodes | | **Load Asset** | Creates an Asset node to pick existing media | | **Upload…** | Uploads files from your computer as Asset nodes | | **Recent nodes**, **Favourite nodes** | Nodes you used recently or starred | | **Latest models** | Newly added models, grouped by media type | | **Image**, **Video**, **Audio**, **Text** sections | Tasks such as generating or editing, each with a Creative Tool option (**Any model**) and a short list of models, plus **Browse all N models…** when there are more | Right-click an input or output socket to get the same kind of menu, limited to nodes that can connect to that socket. ## Add a model from the model hub Press `Shift+M` to show or hide the model hub, a panel of models sorted into **Image**, **Video**, **Audio**, **Text** and **3D** tabs and task groups such as **Generate**, **Edit** and **Upscale**. Click a model to add it to the canvas, drag it to place it, or double-click to add it and close the hub. If the selected node's output does not fit the model, the hub shows **Not compatible with current selection**. For help choosing, see [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). ## Add a node from the Node Browser The Node Browser lives in the **Library** panel under **Nodes**. Use it when you want to read about a node before adding it: inputs, outputs, parameters, examples and pricing information. - Drag a node card onto the canvas to create it there. - Right-click a card and choose **Create Node to Graph**. Open the Library panel from **Window** > **New Panel Window** > **Library**. ## Add media from the Asset Manager or your computer - Drag an asset from the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/) onto the canvas to create an Asset node. - Drag files from your computer onto the canvas. They upload and become Asset nodes automatically. - Press `O` to open **Add asset**, then choose **Assets**, **Upload** or **Load from URL**. Upload types and limits are listed on [Upload and organize assets](https://docs.nodespell.com/assets/upload-and-organize/). ## Copy, paste and duplicate nodes | To | Do this | | ----------------------- | ---------------------------------------- | | Copy the selected nodes | `Cmd/Ctrl+C` | | Cut | `Cmd/Ctrl+X` | | Paste | `Cmd/Ctrl+V` | | Duplicate by dragging | Hold `Option/Alt` and drag the selection | | Delete | `Backspace` or `Delete` | Copied nodes keep their parameters and the connections between them. Copying a Folder includes every node inside it. Deleting a node that sits in the middle of a chain reconnects the node before it to the node after it, so the chain is not broken. ## Frequently asked questions ### What does the Tab menu search? The Tab menu searches node names, models, Creative Tools, custom nodes and workflows together. Typing in Chinese also matches pinyin. ### Why is a node missing from the Tab menu? Data, logic and math nodes are hidden until you turn on **Advanced nodes** in the menu footer. A **Filter** or **Input** filter can also hide results; clear them to see everything. ### Why does my new node connect automatically? When exactly one node is selected, the Tab menu creates the new node beside it and connects the first compatible socket. Press `Esc` to clear the selection first if you want an unconnected node. ### Does previewing an effect add it to my graph? No. The Tab menu renders the effect on the selected image for comparison only. Nothing is added until you press `Enter` or click the result, and previews are shown only for realtime image effects, not for AI models. ## Related - [Navigate the Graph Editor](https://docs.nodespell.com/graph/graph-editor/): pan, zoom and select - [Connect nodes](https://docs.nodespell.com/graph/connections/): what connects to what - [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/): reuse saved setups and defaults - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/): pick the right model node - [Asset Manager](https://docs.nodespell.com/assets/asset-manager/): find media to bring into the graph --- # Connect nodes and manage connections > Connect nodes in Nodespell by dragging between sockets, learn which content types fit, and reroute, reorder, cut or replace connections without breaking the graph. Source: https://docs.nodespell.com/graph/connections/ · Last updated: 2026-11-10 A connection in Nodespell is a link from one node's output socket to another node's input socket. Connections carry text, images, video, audio or data downstream, so a prompt can feed a model and a model's result can feed an effect. Each input checks what it receives and tells you when the content type does not match. ## Connect two nodes 1. Press on an output socket of the source node. In the default top-to-bottom flow, outputs are on the bottom edge. 2. Drag towards the target node. The connection snaps to the nearest suitable input socket and shows its name. 3. Release to create the connection. You can also drag from an input socket back to an output. Hover any socket for a moment to see its name, whether it is an input or output, its type, and **Multiple** if it accepts several connections. To add a node that is already connected, right-click a socket and pick a node from the menu, which only lists nodes that fit that socket. Selecting a node and pressing `Tab` also connects the node you create. ## What can connect to what? Every socket has a content type and a data type. The content type is what matters day to day. | Content type | Carries | | ------------ | ------------------------------------------------------ | | **Text** | Prompts, captions and other text | | **Image** | Still images | | **Video** | Video clips | | **Audio** | Speech, music and sound | | **3D model** | 3D meshes | | **Any** | Whatever is connected; used by general-purpose sockets | Matching content types connect cleanly. A socket of type **Any**, or a data type of **Inferred**, adapts to what it receives. If you connect different media types, for example an audio output into a video input, Nodespell still creates the connection but labels it **Expects Video, got Audio** while you drag. Before a run that includes it, a **Check input connections** dialog lists each mismatch; choose **Cancel** to fix it or **Run anyway** if the mismatch is intentional. ## How do single and multiple inputs differ? | Input kind | What happens when you connect | | ------------ | --------------------------------------------------------------------------------------------- | | Single input | Takes one connection. Connecting a new source replaces the old connection. | | Multi-input | Takes several connections, up to the limit the node sets. The socket info shows **Multiple**. | A multi-input's connections have an order, which matters for nodes such as reference-image inputs or merges. In Parameters, a multi-input shows its connection count and three ordering buttons: - **Order By Position**: orders connections by where their source nodes sit on the canvas. - **Order By Name**: orders them alphabetically by source node name. - **Reverse**: reverses the current order. The [Node Browser](https://docs.nodespell.com/graph/add-nodes/) lists each input's minimum and maximum number of connections. ## Move, reroute and replace connections | To | Do this | | ---------------------------------------------- | ------------------------------------------------------- | | Move one connection to another socket | Drag the end of the connection line onto the new socket | | Move all connections on a socket | Drag the socket itself onto another socket | | Reroute a connection through another node | Drag the middle of the connection onto that node | | Insert a node into an existing connection | Drag the node onto the connection | | Swap the first two inputs of the selected node | `Shift+R` | Dropping onto an occupied single input replaces whatever was connected there. ## Tidy long connections with Dots A Dot is a tiny pass-through node that bends a connection or splits it to several targets. - Press `.` (period) to create a Dot at the pointer. If one node is selected, the Dot is created connected to it. - Right-click a connection and choose **Add Dot**. - Hold `Option/Alt` and drag from a connection to pull out a Dot where you release. Double-click a Dot to open **Edit Dot Text** and label it. ## Disconnect and cut connections | To | Do this | | ---------------------------------- | ------------------------------------------------------------ | | Remove one connection | Double-click it, or right-click it and choose **Disconnect** | | Cut several connections | Hold `Y` and drag across them, then release | | Pull selected nodes out of a chain | Press `X`, or shake a node while dragging it | | Delete a node but keep the chain | Select it and press `Backspace` or `Delete` | `X`, shaking and deleting all reconnect the nodes on either side so the rest of the chain keeps flowing. Every change can be undone with `Cmd/Ctrl+Z`. ## Connect an input to a parameter Many parameters can be driven by a connection instead of a typed value. Drag a connection onto a node; if the node has parameters that can accept it, a **More inputs** target appears. Click it to open **Connect to hidden input**, which lists **Best matches** and **Other eligible parameters**. You can also right-click a parameter in Parameters and choose **Connect as input**. See [Drive a parameter from a connection](https://docs.nodespell.com/graph/parameters/). ## Why is a connection rejected? When Nodespell cannot create a connection, it leaves the graph as it was. The usual reasons: | Reason | What to do | | ------------------------------------------------------------- | ---------------------------------------------------------------------------------------- | | The same connection already exists | Nothing; the link is already there | | The multi-input already has its maximum number of connections | Disconnect one first, or use another node | | Several sources were dropped on an input that takes one | Connect one source, or use a multi-input (**Requires multi-input to connect N sources**) | | The socket no longer exists, for example after a node update | Reconnect to the node's current sockets | Inside a compositor, some sockets accept only compositor-compatible sources. See [Compositing](https://docs.nodespell.com/edit/compositing/). ## Frequently asked questions ### Can I connect an image output to a video input? Yes, but Nodespell marks it as a mismatch and asks you to confirm with **Run anyway** before running. Most models ignore or misread media of the wrong type, so check the connection first. ### What happens to the old connection when I connect a new one to the same input? On a single input, the new connection replaces the old one. On a multi-input, the new connection is added alongside the existing ones until the input's limit is reached. ### How do I change the order of reference images on a multi-input? Select the node and use **Order By Position**, **Order By Name** or **Reverse** on that input in Parameters. **Order By Position** follows the left-to-right position of the source nodes. ### Can I cut connections with a trackpad? Yes. Hold `Y` and drag across the connections with a normal trackpad click, or double-click a single connection to remove it. ## Related - [Navigate the Graph Editor](https://docs.nodespell.com/graph/graph-editor/): sockets and node parts - [Add nodes](https://docs.nodespell.com/graph/add-nodes/): create nodes already connected - [Edit parameters](https://docs.nodespell.com/graph/parameters/): drive parameters from connections - [Run nodes](https://docs.nodespell.com/graph/run-nodes/): what runs when connections change - [Organize your graph](https://docs.nodespell.com/graph/organize/): groups and tidy layouts --- # Edit node parameters > Edit node settings in the Nodespell Parameters panel with sliders, number fields and the colour picker, then pin, reset, or save defaults. Source: https://docs.nodespell.com/graph/parameters/ · Last updated: 2026-11-10 Parameters are a node's settings in Nodespell, such as a model's aspect ratio, a blur's size or a prompt's text. You edit them in the **Parameters** panel, which shows the selected node. A parameter can hold a typed value, follow an [expression](https://docs.nodespell.com/graph/expressions/), be animated with [keyframes](https://docs.nodespell.com/edit/keyframes/), or be driven by a connection. ## Open the Parameters panel Select a node and its parameters appear in **Parameters**. When nothing is selected the panel says **No nodes are selected or pinned**. - Press `Q`, or turn on **Docked Parameter Editor** in **View Settings**, to dock Parameters beside the graph. - Open Parameters as its own panel from **Window** > **New Panel Window** > **Parameters**. - Select several nodes of the same kind to edit a shared parameter on all of them at once. ## Keep a node in view with pinning Click **Pin/Unpin Nodes** in the Parameters toolbar to pin the node you are editing. Parameters keeps showing the pinned node while you select other nodes, which is useful when you tweak one node and watch another. Click the button again to unpin. ## Which controls does Parameters use? | Control | Used for | Notes | | ------------------------ | --------------------------------------------- | -------------------------------------------------------------- | | Slider with number field | Decimal and whole numbers | Drag the slider or type in the field | | Toggle | On/off settings | | | Menu | One choice, or several for multi-choice menus | | | Text field | Short text | | | Long text field | Prompts and other long text | Select **Expand** for more room; **Back to parameter** returns | | Colour | Colours and linked RGB values | Swatch opens the colour editor | | Asset picker | Images, video or audio from your assets | | | List | Several values or points | Add, remove and reorder items | | Curve and ramp editors | Colour and value ramps | Used by colour and grading nodes | Some parameters can be cleared with **Unset**, which leaves the value empty so the model uses its own default. ## Fine-tune a number with press and hold The value ladder gives you precise control without a middle mouse button. 1. Press and hold on a number field or a slider's handle. A ring fills around the pointer; keep holding until the ladder opens. 2. Without releasing, move up or down to choose a step size: **100**, **10**, **1**, **0.1**, **0.01** and smaller. Whole-number parameters stop at **1**. 3. Drag left or right to change the value by that step. The ladder shows the new value and how far it has moved. 4. Release to finish. The whole drag is one undo step. Hold the pointer at the left or right edge of the window to keep the value moving. With a mouse, holding the middle button opens the ladder straight away. ## Type exact numbers and calculations - A slider's ends are only its travel range. You can type, drag or nudge a number field past them. - Type a calculation such as `1920 / 2` and Nodespell stores the result as a fixed number. - With several nodes selected, type a range such as `0..100` to spread values evenly across them in selection order, or `0..100,0.6` to set the curve. **Distribute Values…** in the parameter's right-click menu does the same with a dialog. Keyboard nudges in a number field: | Keys | Change | | -------------------------------------- | ------------------------ | | `Shift+Up` / `Shift+Down` | Add or subtract 10 | | `Option/Alt+Up` / `Option/Alt+Down` | Add or subtract 0.1 | | `Option/Alt+Right` / `Option/Alt+Left` | Multiply or divide by 10 | | `Cmd+Up` / `Cmd+Down` (macOS) | Add or subtract 1 | | `Cmd+Right` / `Cmd+Left` (macOS) | Multiply or divide by 2 | When a slider drag lands on the parameter's default, it snaps there briefly; hold `Option/Alt` to move freely. ## Pick colours precisely Click a colour swatch to open the colour editor. - Drag in the saturation and value square or the hue strip. Hold `Shift` or `Option/Alt` to move a tenth as far, or both for a hundredth. - Type exact values in RGB or switch to HSV with the mode button. HSV needs values of 0 or more. - **Intensity** multiplies the colour so values above 1, used for HDR work, stay editable. - Click the original swatch to revert to the colour you started with. **Reset to default** and **Make neutral** are in the editor too. - Open **Larger colour picker** for more room. In the parameter row, **Show RGB channels** splits the colour into R, G and B fields. Arrow keys nudge a channel; hold `Shift` for finer steps or `Cmd/Ctrl` for larger ones. Simple colour fields also offer a hex value, **Save swatch**, and **Pick from screen** in browsers that support it. ## Reset a parameter or use your saved defaults | To | Do this | | ------------------------------------------------ | --------------------------------------------------------------------------------------------- | | Reset one parameter | Double-click its label, `Option/Alt`+click it, or right-click and choose **Reset to default** | | Reset every parameter on the node | **Revert Parameters to Default** in the toolbar's **More** menu | | Save the node's current settings as your default | Right-click the node on the canvas and choose **Save as my default…** | | Manage your saved defaults | Preferences > **My node defaults** | A saved default applies to nodes of that type you add from now on, in any project; existing nodes do not change. Once a default is saved, **Reset to default** restores your saved value instead of the Nodespell value. Values that come from a connection, an expression, keyframes, project files or references cannot be saved. Resetting a parameter that has an expression removes the expression, with an **Undo** button in the message. See [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/) for named presets. ## Drive a parameter from a connection Right-click a parameter and choose **Connect as input** to add an input socket for it, then connect a node to that socket. You can also drag a connection onto the node and choose the parameter from **Connect to hidden input**. While connected, the row shows **From** and the source node's name, and the control is read-only: **This parameter is controlled by an input connection from** that node. Choose **Disconnect input** to return to the typed value. When more than one thing could set a parameter, Nodespell uses the first that applies: a connection, then keyframes, then an expression, then the typed value, then the default. ## Show a value on the node Right-click a parameter and open **Show on the node** to display its value on the node in the graph: **Evaluated** shows the computed value, the raw option shows the typed value or expression text, and **No Display** hides it. `Cmd/Ctrl`+click a parameter label to cycle through these. ## Frequently asked questions ### Can I type a value larger than the slider allows? Yes. The slider's range only sets how far the handle travels; the number field accepts values beyond it. ### How do I fine-tune a value on a trackpad? Press and hold the number field or slider handle until the value ladder opens, move up or down to pick a step size, then drag sideways. No middle button is needed. ### Why can't I edit a parameter? A parameter is read-only when a connection drives it or when the project is open for viewing only. Choose **Disconnect input** to type a value again. ### Do saved defaults change nodes I already have? No. A saved default applies only to nodes of that type you add afterwards. To bring an existing node to your default, use **Reset to default** on its parameters. ## Related - [Expressions](https://docs.nodespell.com/graph/expressions/): drive a parameter with a calculation - [Connect nodes](https://docs.nodespell.com/graph/connections/): sockets and multi-inputs - [Keyframes](https://docs.nodespell.com/edit/keyframes/): animate parameters over time - [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/): reusable setups - [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/): edit long prompt fields --- # Drive parameters with expressions > Write Nodespell expressions that calculate a parameter from other parameters, nodes, time and functions, with syntax, references, a function table and error fixes. Source: https://docs.nodespell.com/graph/expressions/ · Last updated: 2026-11-10 An expression in Nodespell is a short calculation that sets a parameter's value, such as `blur1.size * 2` or `if($F < 24, 0, 1)`. Expressions link parameters together, so changing one value updates every parameter that reads it, and in compositors they can change over time. The parameter shows the live result in its normal control. ## Add an expression to a parameter 1. Select the node and find the parameter in **Parameters**. 2. Right-click the parameter and choose **Use an expression**. In a text or number field you can also type `=` when the field is empty or all of its text is selected. 3. Type in the expression bar. Suggestions list **Functions**, **This node**, **Nodes**, **Parameters** and **Time**; press `Tab` or `Enter` to accept one. 4. Read the result line under the bar as you type. 5. Press `Enter` or `Cmd/Ctrl+Enter`, or select **Apply**, to save. Press `Esc` to cancel. Select **Pick a parameter** in the bar, then click a parameter in Parameters or a node on the canvas, to insert a reference without typing it. **Functions** opens a searchable list of every function with examples. Empty expressions offer starters such as **Use this node's own values**, **Switch on a condition**, **Keep a value in range** and **Change over time**. To edit later, right-click the parameter and choose **Edit expression…**. Hover the expression chip to see its **Result**, **Type** and what it **Reads**, with **Edit**, **Bake to value** (replace the expression with its current result) and **Remove expression**. **Show expressions** in Parameters switches every row to show its source text instead of its result. ## Refer to other parameters and nodes | To read | Write | | --------------------------------------------- | ------------------------------------- | | A parameter on this node | `$size` | | A parameter on another node on the same Level | `blur1.size` | | The node this Level is inside | `parent.size` | | A node one Level further out | `parent.parent.blur1.size` | | A node inside a Group on this Level | `group1.blur1.size` | | A node from the top of the project | `root.tex.blur1.size` | | One channel of a colour | `grade1.gain.r`, `$gain.g` | | A name with spaces or symbols | `"my blur".size`, `blur1."edge size"` | Names are case-sensitive. A reference reads the other parameter's current value: if that parameter has its own expression, its result; if it is animated, its value at that node's frame. A parameter driven by a connection cannot be read, because its value is only known when the graph runs. ## Time variables In compositor nodes, five variables describe the current frame. They are not available elsewhere. | Variable | Meaning | | -------- | -------------------------------------- | | `$F` | Current frame | | `$T` | Current time in seconds | | `$PERC` | Progress through the clip, from 0 to 1 | | `$DURT` | Duration in seconds | | `$DURF` | Duration in frames | ## Values and operators Expressions work with numbers, text, true/false, colours and lists. Constants are `true`, `false` and `pi`. | Operators | Meaning | | --------------------------------- | ------------------------------------------------------------------ | | `+ - * /` | Arithmetic. `+` joins text when either side is text. | | `%` | Remainder, always with the sign of the divisor, so `-1 % 4` is `3` | | `^` | Power | | `== != < > <= >=` | Comparison | | `and`, `or`, `not` (or `&&`, `!`) | Logic; `or` can also be written as two vertical bars | | `[1, 2, 3]`, `list[0]` | Lists and indexing; a negative index counts from the end | Text uses double or single quotes. Colours support `+` and `-` with another colour, and `*` and `/` with a number, per channel. A single `=` is an error; use `==` to compare. ## Functions Function names are lowercase; the editor fixes the case for you. | Group | Functions | | ---------------- | ------------------------------------------------------------------------------------------------------------------------------------------ | | Maths | `abs` `sign` `floor` `ceil` `round(x, digits)` `trunc` `sqrt` `pow` `exp` `ln` `log(x, base)` `mod` `min` `max` `clamp(x, lo, hi)` | | Angles (radians) | `sin` `cos` `tan` `asin` `acos` `atan` `atan2(y, x)` `degrees` `radians` | | Shaping | `lerp(a, b, t)` `fit(x, inLo, inHi, outLo, outHi)` `smoothstep(lo, hi, x)` `rand(seed)` | | Logic | `if(condition, a, b)` | | Text | `upper` `lower` `trim` `length` `replace(s, find, with)` `contains(s, part)` `split(s, sep)` `join(list, sep)` `pad(n, width)` `number(s)` | | Lists | `range(n)` or `range(start, end)` `first` `last` `sum` `mean` `sort` `reverse` | | Colour | `rgb(r, g, b)` | Useful details: `round` rounds halves away from zero; `fit` stops at the ends of the new range but `lerp` does not; `rand` returns a number from 0 up to 1 and the same seed always gives the same number; `range(3)` is `[0, 1, 2]`; `pad(3, 3)` is `"003"`; `rgb` values above 1 are allowed. ## Worked examples | Goal | Expression | | --------------------------------------- | ----------------------------------- | | Keep a second blur at twice the first | `blur1.size * 2` | | Match this node's height to its width | `$width` | | Fade in over the first second of a clip | `clamp($T, 0, 1)` | | Pulse a value once per 24 frames | `sin($F / 24 * 2 * pi) * 0.5 + 0.5` | | Switch at frame 48 | `if($F < 48, 0, 1)` | | Number shot names | `"shot_" + pad($F, 4)` | | Use only the red channel of a grade | `grade1.gain.r` | | A fixed random value per seed | `rand("hero") * 10` | ## Fix expression errors When an expression has a problem, the parameter uses its default and the row shows the message. A spelling or reference problem anywhere in the expression stops it from running, even in an `if` branch that would not be taken. | Message | Cause and fix | | ---------------------------------------------------- | ---------------------------------------------------------------------------------------- | | **"…" is not a node, a function or a value.** | A typo or missing node. Use the suggested name if one is offered. | | **There is no node named … on this level.** | References look on the current Level. Use `parent.` or `root.` for other Levels. | | **That node has no parameter named ….** | Check the parameter name; hover it in Parameters to see it. | | **Wrong number of values.** | The function takes a different number of arguments; the message shows how it is written. | | **This expression ends up reading itself.** | Two expressions read each other. Break the loop. | | **$F only works on compositor nodes.** | Time variables are only available in compositors. | | **This reads a parameter that a connection drives.** | Read a typed value instead, or connect the same source to this node. | | **This divides by zero.** | Guard the divisor, for example with `max(x, 0.0001)`. | | **The result is …, but this parameter needs ….** | Convert the result, for example with `number()` or `rgb()`. | Expressions also have limits: 4,096 characters, 64 levels of nesting, lists of 10,000 items and text of 65,536 characters. ## Frequently asked questions ### Which wins if a parameter has an expression and a connection? The connection wins, then keyframes, then the expression. An expression that is not in use stays saved and editable, and the row says what is driving the parameter. ### How do I turn an expression back into a normal value? Hover the expression and choose **Bake to value** to keep the current result as a fixed value, or **Remove expression** to drop it. **Use a fixed value** in the right-click menu does the same as baking. ### Can I use expressions outside compositors? Yes. References, operators and functions work in any node's parameters. Only the time variables need a compositor node. ### Does Reset to default keep my expression? No. Resetting a parameter removes its expression; the message offers **Undo** if you change your mind. ## Related - [Edit parameters](https://docs.nodespell.com/graph/parameters/): typed values, connections and defaults - [Keyframes](https://docs.nodespell.com/edit/keyframes/): animate values on the timeline - [Compositing](https://docs.nodespell.com/edit/compositing/): where time variables apply - [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/): how `parent` and `root` paths work - [Data and logic nodes](https://docs.nodespell.com/graph/data-and-logic/): calculations as nodes instead of expressions --- # Write prompts and reuse saved text > Write prompts in Nodespell with the Prompt node, insert smart literals, reuse Saved text with /, and dictate with voice typing. Source: https://docs.nodespell.com/graph/prompts-and-saved-text/ · Last updated: 2026-11-10 A Prompt node in Nodespell holds text that you connect to one or more models, so the same prompt can drive several generations. The prompt editor adds tools for writing it, and the Saved text library keeps prompts and phrases you reuse, ready to insert into any text field by typing `/`. ## Create and edit a Prompt node 1. Point at the canvas and press `P`, or right-click empty canvas and choose **Prompt**. 2. Type straight into the node. Press `Enter` on a selected Prompt node later to edit it in place, and select **Done** when finished. If the text is longer than the node, **Fit node** makes the node tall enough. 3. For more room, right-click the node and choose **Edit Prompt**, or select **Open in full editor** while editing in place. 4. Connect the Prompt node's output to a model's prompt input. See [Connect nodes](https://docs.nodespell.com/graph/connections/). Prompt fields on model nodes in **Parameters** use the same writing tools as the Prompt node. ## What can the prompt editor do? The full prompt editor opens in its own dialog. Press `Cmd/Ctrl+Enter` to save or select **Save**; **Cancel** asks before discarding changes. | Tool | What it does | | ------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Add** (`/`) | Opens a menu of saved text, smart literals and, on relighting models, **Lighting** | | `@` references | When the prompt feeds a node that accepts reference media, type `@` to reference connected media; the placeholder says **Press / to add, @ to reference** | | **Edit with AI** | Rewrites the **Whole prompt** or the **Selection**: **Shorten**, **Expand**, **Translate**, **Style**, **JSON**, **Safer rewrite**, or a **Custom instruction**. Review **Changes** or **Result**, then **Apply** | | **Assistant** | Opens a chat beside the prompt that drafts and rewrites it. Nothing is saved until you select **Save** | | **Sensitive words** | In the view options, highlights words that may trigger content filters, graded **Likely** or **Maybe**. English only, and a guide rather than a guarantee | | **Text size** | Small, Default, Large or Extra large | | **Voice typing** | Dictates into the prompt; see below | If a prompt is computed by an [expression](https://docs.nodespell.com/graph/expressions/), the editor says so and you change it in Parameters instead. ## Insert smart literals Smart literals are values in a prompt that Nodespell recognises and lets you adjust with a small control: a hex colour such as `#FF7A3D`, a measurement such as `1.8m` or `35mm`, an angle such as `45°`, and an aspect ratio such as `16:9`. Type them straight into the prompt, or insert one from **Add** > **Smart literals**. Click a smart literal to change its value or unit, pick a colour, or choose another aspect ratio. ## Reuse text from your Saved text library Saved text is your personal library of prompts and phrases. It belongs to your account and follows you to every project and device. 1. Click in any text field, including a prompt, at the start or after a space. 2. Type `/`. A list of your saved text opens; keep typing to search it. 3. Use the arrow keys to choose and press `Enter` to insert, or `Esc` to close. You can also right-click a text field and choose **Insert saved text**. In a single-line field, line breaks in the saved text become spaces. To save new text, select it in a text field, right-click, and choose **Save selection as text…**. Give it an optional name and select **Save**. ## Manage saved text in Preferences Open Preferences with `Cmd/Ctrl+,` and go to **Saved text**, or choose **Manage saved text…** from a text field's right-click menu. - Search names and text, and sort by **Name**, **Recently used on this device** or **Recently edited**. - Select an entry to edit its name and text, then **Save**. - **Pin to top** keeps an entry first in the list; **Delete** removes it from all your devices. - Entries that need a decision appear under **Needs attention**, for example a **Conflict** when the same text was changed on another device. Choose **Keep mine** or **Use other version**. | Limit | Value | | -------------- | ----------------- | | Text per entry | 20,000 characters | | Name | 200 characters | When you are offline, the list shows only saved text already on this device. ## Dictate with voice typing Select the microphone button labelled **Voice typing** in the prompt editor or under a long text field, allow microphone access, speak, then select **Finish**. The transcript is inserted at the cursor. Turn on **Clean up speech** to remove fillers and repetition while keeping your language. Voice typing **Uses credits**, charged by processing cost, and cleanup adds a small extra charge when it succeeds. It needs a supported browser and microphone access. ## Frequently asked questions ### Why doesn't typing / open my saved text? The `/` list opens only at the start of a field or after a space, bracket or quote. Typing `/` in the middle of a word, such as in a URL, inserts a normal slash. ### Is saved text shared with my Team? No. Saved text is personal to your account. It syncs across your devices but is not visible to other people. ### Does Edit with AI change my prompt straight away? No. Edit with AI shows the proposed change first. Nothing changes until you select **Apply**, and the prompt editor still needs **Save**. ### Can I use one prompt for several models? Yes. Connect a Prompt node's output to the prompt input of each model. Editing the prompt updates every connected model the next time it runs. ## Related - [Connect nodes](https://docs.nodespell.com/graph/connections/): wire a prompt into models - [Edit parameters](https://docs.nodespell.com/graph/parameters/): prompt fields on model nodes - [Expressions](https://docs.nodespell.com/graph/expressions/): compute text from other values - [Account settings](https://docs.nodespell.com/account/settings/): Preferences, including Saved text - [Relight](https://docs.nodespell.com/creative-tools/relight/): models that offer the Lighting control --- # Organize your graph with groups, tidy and find > Keep large Nodespell graphs readable with nested Groups, Tidy up and Layout mode, node colours, names and comments, and Find in graph to search and replace. Source: https://docs.nodespell.com/graph/organize/ · Last updated: 2026-11-10 Nodespell gives you several ways to keep a graph readable as it grows. Groups frame related nodes, Tidy up and Layout mode arrange nodes for you, colours, names and comments label them, and Find in graph searches every node by name, comment or parameter. None of these change what the graph produces. ## Group related nodes A Group is a coloured frame with a title that holds nodes together on the canvas. Moving a Group moves everything inside it. 1. Select the nodes you want to group. 2. Press `Cmd/Ctrl+G`, or right-click one of them and choose **Group Selected Nodes**. 3. Name the Group with **Rename** in the **Group options** menu in its header, and use the colour button in the header to pick a colour. - Drag a node into a Group to add it; drag it out to remove it. - Groups can be nested: drag a Group into a larger Group to organise in layers. A Group can only drop into a Group that is bigger than it. - The **Group options** menu in the header has **Rename**, **Fit to contents** (shrink or grow the frame to its nodes), **Select contents** (select the Group and everything nested in it) and **Ungroup**. - **Ungroup** removes the frame but keeps its nodes. In a nested Group, the nodes move up into the Group around it. Groups are visual only; they do not change how nodes run. To hide part of a graph inside a single node, use a [Level or Folder](https://docs.nodespell.com/graph/levels-and-folders/) instead. ## Tidy up a messy graph Press `L`, or choose **Layout** > **Tidy up** in the top bar. Tidy up arranges the selected nodes, or the whole Level if nothing is selected, with clearer branches and spacing that fits your nodes. Groups keep their contents, and separate clusters of connected nodes (islands) are spaced apart. A message confirms how many nodes moved, for example **Tidied 9 of 12 nodes**, with **Undo** and **Adjust…**. The whole tidy is one undo step. If nothing needs to move, the message says **Already tidy**. ## Fine-tune layout in Layout mode Layout mode treats the graph as a shape you can reshape while Nodespell keeps nodes from overlapping. Open it with `Shift+L`, **Layout** > **Layout mode**, or **Adjust…** after a tidy. | Control | What it does | | ---------------------------------------------------- | -------------------------------------------------------------------------------------------------------- | | **Auto** dials: **Tidy**, **Islands** | Tidy gradually, from **None** through **Light** and **Medium** to **Full layout**, and separate islands | | **Adjust** dials: **Spacing**, **Tension**, **Size** | Change gaps, straighten chains and scale node sizes; **Axis** limits spacing to along or across the flow | | **Transform** | Rotate 90° left or right, flip horizontally (`H`) or vertically (`V`) | | **Select by relationship** | **Upstream** (`U`), **Downstream** (`D`), **Island** (`I`) | | **Behaviour** | **Push others**, **Allow overlap**, **Pack**, **Scale sizes** | | **Pin** (`P`) | Keeps selected nodes fixed while others move | | **Compare** and **Outlines** | Hold to compare with where you started; show outlines of the starting positions | Arrow keys nudge the selection (`Shift` for bigger steps). `Cmd/Ctrl+Z` undoes one step inside Layout mode. Press `Enter` or select **Done** to keep the result as one undo step, or **Cancel** to restore the original. `Esc` cancels the current drag first, then clears the selection, then asks **Discard N changes?**. ## Colour, rename and comment nodes Right-click a node to label it: | Item | Effect | | ---------------- | ----------------------------------------------------------------------------------------------------------------- | | **Node Colour** | Sets the colour of the node, or of every selected node, from common colours, **More colours**, or a custom colour | | **Rename Node** | Renames the node. With several selected, enter one base name and Nodespell adds suffixes to keep names unique | | **Edit Comment** | Adds a note saved with the node. Comments show on the canvas when **Node Comment** is on in **View Settings** | In light mode, node colours are lightened on the canvas so nodes stay readable; your chosen colour is unchanged. Dots can carry short labels too; see [Connect nodes](https://docs.nodespell.com/graph/connections/). ## Find nodes with Find in graph 1. Press `Cmd/Ctrl+F` to open **Find in graph**. 2. Type part of a node name, a comment or a parameter value. Name matching is forgiving, so `hrpr` can find **Hero Prompt**. 3. Use the arrow keys to move through matches and press `Enter`, or click **Go to**, to jump to a node. Nodespell opens the node's Level and frames it. | Option | Choices | | ------------------- | --------------------------------------------------------------------- | | Inside the field | **Match case**, **Whole word**, **Regular expression** | | **Where to search** | **Project**, **This level** (and the Levels inside it), **Selection** | | **What to search** | **Everything**, **Names**, **Comments**, **Parameters** | Up to 400 matches are listed; narrow the search to see the rest. In a large project, project-wide search covers names and folder paths straight away; select **Load full project** to include comments and parameters. ## Replace text across the graph Press `Cmd/Ctrl+Shift+F`, or select **Replace** in Find in graph. Every match shows the text it will become. Untick any match you want to keep, then press `Enter` in the replace field to replace the highlighted match, or `Cmd/Ctrl+Enter` to replace every ticked match. A replace-all is a single undo step. Numbers and on/off values are not replaced, and prompts that contain references must be edited in the prompt editor. Expressions are searched and replaced in their expression text. ## Frequently asked questions ### Does grouping nodes change how they run? No. Groups only frame nodes on the canvas. Use a Level when you want a self-contained piece of graph with its own inputs and outputs. ### Can I undo Tidy up? Yes. Each tidy is one undo step: press `Cmd/Ctrl+Z` or select **Undo** in the message. ### Does Find in graph search other Levels? Yes. Set **Where to search** to **Project** to search every Level, or **This level** for the current Level and those inside it. ### What is the difference between Tidy up and Layout mode? Tidy up arranges the selection or Level in one step at full strength. Layout mode lets you dial the amount, reshape, pin and push nodes by hand, and keeps all of it as one undo step when you select **Done**. ## Related - [Navigate the Graph Editor](https://docs.nodespell.com/graph/graph-editor/): select, align and frame nodes - [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/): nest parts of the graph - [Connect nodes](https://docs.nodespell.com/graph/connections/): Dots and tidy connections - [Keyboard shortcuts](https://docs.nodespell.com/reference/shortcuts/): every layout key --- # Organize graphs with Levels and Folders > Nest parts of a Nodespell graph inside Levels and Folders, move in and out with the breadcrumb, set flow direction per Level, and open large projects faster. Source: https://docs.nodespell.com/graph/levels-and-folders/ · Last updated: 2026-11-10 A Level in Nodespell is a node that contains its own graph, so a whole section of work can live inside one node. A Folder is a Level set to organise rather than compute: it holds nodes, shows a cover, and cannot be run or connected. Both keep large projects tidy, and you move in and out of them like folders on a computer. ## What is the difference between a Level and a Folder? | | Level | Folder | | --------------------------- | ---------------------------------- | ------------------------------------------------------------------------ | | Purpose | A self-contained part of the graph | Organisation, like a folder of files | | Sockets in the parent graph | Yes; can connect to other nodes | No sockets | | Can run | Yes, as part of the graph | No: **Folders cannot be run directly.** | | On the canvas | A node | A folder card with an item count, optional **Cover Image** and **Title** | | Convert | **Convert to Folder** | **Convert to Level** | Both are the same kind of node, so you can switch between them. A Level that has connections in its parent graph cannot become a Folder until you remove them: **Disconnect parent-level connections before converting this level to a folder.** ## Create a Level or Folder - Right-click empty canvas and choose **Folder**. - Press `Tab` and search for **Level** to add a Level. - Right-click a Level node and choose **Convert to Folder**, or right-click a Folder and choose **Convert to Level**. The same commands are in the command palette (`Cmd/Ctrl+K`). Set a Folder's **Cover Image** and **Title** in **Parameters** with the Folder selected. ## Go into and out of a Level or Folder | To | Do this | | ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------- | | Go in | Double-click the node, select it and press `Enter`, or right-click it and choose **Enter Node** (Levels) or **Open Folder** (Folders) | | Go up one Level | Press `U`, click the up arrow (**Go to parent level**) in the Graph Editor bar, or right-click empty canvas and choose **Go Up** | | Jump to any outer Level | Click its name in the breadcrumb at the top of the Graph Editor; the first item is **Root** | Nodespell remembers where you were looking on each Level, so returning to a Level brings back its view. ## Stay on one Level while you work By default, selecting a node that lives on another Level, for example from [Find in graph](https://docs.nodespell.com/graph/organize/), takes the Graph Editor to that node's Level. Click the lock button, **Pin/Unpin current level**, next to the up arrow to keep the Graph Editor on the current Level instead. Click it again to unpin. ## Set the flow direction for each Level Flow direction sets which way connections run: **Top to bottom**, **Bottom to top**, **Left to right** or **Right to left**. It changes which edges sockets sit on, not the order things run in. 1. Open the Level you want to change. 2. In the top bar, choose **Layout** > **Flow direction**. 3. Under **This level**, pick a direction, or pick the inherited option (**Project default**, **Compositor default** or **Same as parent**) to follow the setting above it. 4. Under **Project defaults**, set **Graphs** and **Compositors** for every Level that does not have its own setting. Flow direction **Applies to everyone in this project**, and only editors can change it. A Level with its own direction shows a small arrow next to its name in the breadcrumb. Changing direction leaves node positions as they are; press `L` to tidy the Level afterwards. New projects start with the directions set in Preferences, under **Node Graph Editor**: **Graph flow in new projects** and **Compositor flow in new projects**. Both default to **Top to bottom**, and existing projects keep their own. ## Copy and paste Levels and Folders Copy a Level or Folder with `Cmd/Ctrl+C` and paste it with `Cmd/Ctrl+V`, like any node. The copy includes every node inside it, however deeply nested. ## How do large projects load? Nodespell does not load every node in a large project at once. A project prepares once, then loads each Level as you go into it, so opening the project and moving between Folders stays fast. Some actions need the whole project. In [Find in graph](https://docs.nodespell.com/graph/organize/), project-wide search covers names and folder paths straight away; select **Load full project** to include comments and parameters, or to replace text across every Level. ## Frequently asked questions ### Can a Folder be part of a run? No. Folders are for organisation only and cannot be run directly. Convert it to a Level with **Convert to Level** if you need it to compute and connect. ### Why can't I convert a Level to a Folder? A Level with connections in its parent graph cannot become a Folder, because Folders have no sockets. Disconnect those connections first, then choose **Convert to Folder**. ### Does each Level have its own flow direction? Each Level can. By default a Level follows the project default for graphs or compositors; pick a direction under **This level** in **Layout** > **Flow direction** to give it its own. ### How do expressions refer to nodes on other Levels? Use `parent.` to read from the Level around the current one and `root.` to start from the top of the project. See [Expressions](https://docs.nodespell.com/graph/expressions/). ## Related - [Organize your graph](https://docs.nodespell.com/graph/organize/): Groups, tidy and Find in graph - [Navigate the Graph Editor](https://docs.nodespell.com/graph/graph-editor/): pan, zoom and frame on each Level - [Expressions](https://docs.nodespell.com/graph/expressions/): `parent` and `root` references - [Compositing](https://docs.nodespell.com/edit/compositing/): compositor flow defaults - [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/): how project changes are saved --- # Run nodes and see what will happen first > Run selected nodes in Nodespell, preview which nodes reuse, update or generate, check the credit estimate, generate several takes, and stop a run. Source: https://docs.nodespell.com/graph/run-nodes/ · Last updated: 2026-11-10 Running nodes in Nodespell executes the selected part of your graph and everything it needs upstream. Before anything starts, Nodespell prepares a run plan that says which nodes reuse their existing [takes](https://docs.nodespell.com/graph/takes/), which update for free, and which generate new results, so you only spend [credits](https://docs.nodespell.com/account/plans-and-credits/) where you choose to. ## Run the selected nodes 1. In **Graph**, select one or more nodes. 2. Look at the run control that appears in the graph toolbar next to the selection count. Its label tells you what will happen (see the table below). 3. Hover the run control to preview the plan on the canvas, then click it, or press `Cmd+Shift+R` on macOS or `Ctrl+Shift+R` on Windows and Linux. The same control appears at the top of the menu when you right-click a node, and **Run Selected Tasks** is available in the command palette (`Cmd/Ctrl+K`). | Label on the run control | What it does | | ------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------- | | **Run** | Uses the takes you already have. Nothing is charged. | | **Generate** or **Generate** with a number | Makes new takes for the selected AI nodes with the current settings, even if nothing changed. Other paid nodes keep their takes. | | **Generate** a number **& run** | Some upstream nodes have no take yet. This generates them, then runs. | | **Up to date** | Everything selected is up to date, so clicking does nothing. | When the run control says **Generate**, its credit estimate sits beside the label. A run that uses no credits shows no price. ## What runs, what is reused and what regenerates? Nodespell reuses paid AI results until you explicitly ask for new ones. Selecting a downstream node and choosing **Run** does not regenerate an upstream image or video that already has a take; it uses the take that node currently uses. Non-AI nodes such as compositing or text nodes recalculate when their inputs change, and that work is free. Each node in the plan gets one of these outcomes: | Outcome | Meaning | | ------------------------------- | ------------------------------------------------- | | **Generate** | Makes a new take and spends credits. | | **Generate first take** | The node has no take yet, so the run creates one. | | **Reuse** | Uses the node's existing take. | | **Keep selected take** | Uses the take you chose to keep for this node. | | **Update · free** | Recalculates a non-AI node at no cost. | | **Up to date** / **Not needed** | Nothing to do for this node. | | **Needs a take** / **Blocked** | The run cannot continue past this node yet. | ## Preview the run plan on hover Hover or keyboard-focus the run control, or any item in its menu, and the canvas shows the plan: each node in the plan is tinted and labelled with its outcome. Spending is shown in the accent color, free updates in green, reuse in a neutral tint, and blockers in amber with a dashed outline. The tooltip repeats the plan as counts under **What happens to each node** and tells you how many planned nodes are off-screen or inside another [Level](https://docs.nodespell.com/graph/levels-and-folders/), so nothing hidden runs without you knowing. ## Choose exactly which nodes generate 1. Click the chevron on the run control (**Run options**) and choose **Plan this run…**. 2. Review **Nodes in this run**. For each AI node you can switch between **Reuse** and **Generate**. **Reuse** is unavailable for a node that has no take yet. 3. Click the button at the bottom. Its label and price update to match your choices. Your choices apply to this run only. When the main action would generate, **Run options** also offers **Run with existing takes**, with its own estimate: it runs with the takes you already have and never generates anything. ## Generate several takes at once Use **Run multiple times…** to run one AI node several times with the same settings, for example to get a choice of images from one prompt. 1. Select a single AI model node. 2. Open **Run options** and choose **Run multiple times…**. 3. Pick the number of runs with the tiles **1** to **8** or the stepper. The dialog shows the **Estimated total** and the price per run. 4. Click **Run** with the count, for example **Run 4 times**. Each run is submitted separately and becomes its own take; some may wait for available capacity. A fixed seed stays unchanged across the runs. All runs reuse the same upstream takes, so if an upstream node has no take yet, prepare it with a single run first. After you submit, the tiles show each run as **Queued**, **Running**, **Complete**, **Failed** or **Cancelled**. ## Check the cost before you run The estimate beside **Generate**, in **Plan this run**, and in **Run multiple times** is shown in credits. It can read as an exact number, a number with **+** when only a minimum is known, **Free**, or **Price unknown**. Live credit estimates are shown on paid plans; on other plans you can still run. If your balance cannot cover the run, the plan says **Not enough credits for this run.** and does not start. > **Note:** Estimates come from the current settings and model. The final charge for each take is shown in its details in the [Takes window](https://docs.nodespell.com/graph/takes/). ## Stop a run While work is in progress, a **running** indicator with a count appears in the graph toolbar. Click it to open **Task queue**, which lists each run with its status, progress and current steps. Click **Stop** to cancel a run. A run that is waiting for this browser shows **Resume**. If a plan is still being prepared, click **Cancel waiting** in the plan dialog. For a batch that could not be confirmed, **Skip the rest…** stops any runs that were not accepted; runs already accepted keep going. See [Tasks panel](https://docs.nodespell.com/graph/tasks/) for history and failures. ## Why is the run control disabled or not running? - **Up to date**: there is nothing to run for the selection. - A run is already starting from this control. Wait for it to finish submitting. - The plan has a blocker, such as **N nodes need a take first**, a take that cannot load, or not enough credits. Clicking opens the plan so you can see why. - The plan could not be prepared. The control changes to **Retry loading plan**, and its tooltip gives the reason, for example a missing media reference in a prompt or an input that is still uploading. - The selection is a Folder. **Folders cannot be run directly.** - You are viewing the project rather than editing it. See [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/). Before running, Nodespell may also show **Check input connections** if an input receives a different media type than it expects. Fix the connection, or choose **Run anyway**. ## Running in projects without takes Takes are turned on per project, and new projects start with them. In a project without takes, the run control is a green **Run** button with a credit estimate, and its chevron (**Task options**) offers **Run multiple times…**, **Restore results** and **Discard results**. Results there are kept as result history rather than numbered takes, and **Lock Result** in the node's **Flags** menu keeps a result for downstream runs. `Cmd/Ctrl+Shift+R` runs the selection in both kinds of project. ## Frequently asked questions ### Does running a downstream node regenerate my images? No. A downstream run reuses the take each upstream AI node currently uses. Choose **Generate** on that node, or switch it to **Generate** in **Plan this run…**, when you want a new result. ### What is the keyboard shortcut to run selected nodes? Press `Cmd+Shift+R` on macOS or `Ctrl+Shift+R` on Windows and Linux. It does the same thing as clicking the run control. ### How many takes can I generate at once? **Run multiple times…** lets you choose from 1 to 8 runs of one AI node. Each run becomes a separate take. ### Why do I not see a credit estimate? Live credit estimates before each run are a paid-plan feature. You can still run nodes on other plans. ### Can I stop a generation after it starts? Yes. Open the **running** indicator in the graph toolbar and click **Stop** on the run. The cancelled take stays in the node's history marked **Cancelled**, with its charge status in its details. ## Related - [Takes](https://docs.nodespell.com/graph/takes/): choose which result the graph uses - [Tasks panel](https://docs.nodespell.com/graph/tasks/): history, failures and credit usage - [Shortcuts](https://docs.nodespell.com/reference/shortcuts/): every keyboard shortcut - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): how credits work - [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/): fixes for runs that will not start --- # Browse and choose takes > Use takes in Nodespell to browse every result a node has made, choose the one your graph uses, keep a favourite or follow the latest, and delete unused takes. Source: https://docs.nodespell.com/graph/takes/ · Last updated: 2026-11-10 Takes are the saved results of a node in Nodespell. Every time a node runs, its output is kept as a numbered take (**Take 1**, **Take 2**, and so on), and you choose which take the rest of the graph uses. Browsing takes never changes anything; only choosing **Use** changes what flows downstream. ## Where can I see a node's takes? | Where | How to open it | What you can do | | ----------------- | -------------------------------------------------------------------------------------- | ------------------------------------------------------- | | On the node | Zoom in and hover a node with results; a chip with the take count appears on its media | Preview takes and use one | | Graph toolbar | Select one node; the takes chip sits beside the run control | Preview takes, use one, open **All takes…** | | Takes window | Right-click the node and choose **Takes…**, or click **All takes…** in the chip | Every action: use, keep, follow latest, delete, details | | Fullscreen viewer | Open the node's media fullscreen; takes run as a rail down the left | Step through takes and use one | The chip's tooltip tells you what the node uses, for example **Uses: Latest (Take 3)**. ## Preview and choose the take the graph uses 1. Open the takes chip on the node or in the toolbar, or open the Takes window. 2. Click a take to preview it. The preview shows **Previewing** and the take name; the output does not change. 3. Click **Use Take** with its number, for example **Use Take 2**. When the graph confirms the change you see **Now uses Take 2.** 4. To stop previewing without changing anything, click **Back to the take in use**, or press `Esc`. Only successful takes can be used. A take that is still generating says **Wait until the take finishes.**, and a failed or cancelled take cannot become the output. Video takes show a poster frame until you click **Play**. ## Keep a favourite or follow the latest take Each node either follows its newest successful take or stays on one take you chose. The **Uses** chip in the Takes window shows which: | Uses | What it means | | ------------------ | ---------------------------------------------------------------------------------------------------------------------- | | **Latest** | Follows the newest successful take. A new run replaces the output. | | **Kept take** | Stays on this take. New takes are added to the history and do not replace it. | | **Mixed takes** | Each output of a multi-output node comes from a different run. Use one take, or **Follow latest**, to make them match. | | **Earlier result** | A result made before take numbers is selected. | Choosing **Use** on an older take keeps that take. To keep the take a node already shows while it follows the latest, open **More take actions** (the **…** button) in the Takes window and choose **Keep this take**. To go back to following new results, choose **Follow latest** in the same menu. ## How do takes affect downstream nodes? Downstream nodes read the take each upstream node currently uses. When you run a downstream node, Nodespell reuses that take instead of regenerating it, so switching to **Take 2** and choosing **Run** downstream rebuilds the rest of the graph from **Take 2** without spending credits on the upstream node. See [Run nodes](https://docs.nodespell.com/graph/run-nodes/) for how the run plan shows **Reuse** and **Keep selected take**. To keep one take available while the node itself moves on, choose **Pin to a branch** in **More take actions**. Nodespell adds a **Pick Result Version** node below the original that always outputs that take. ## Browse takes in the fullscreen viewer In the fullscreen image, video, audio or text viewer, the node's takes appear as a rail of thumbnails, newest first. Press `↑` and `↓` to step through takes, and `←` and `→` to move between selected nodes. The rail's footer shows the take on show and either **In use**, or **Use** and **Back to the take in use**. The viewer's **Takes…** button opens the full Takes window. ## Delete a take 1. Open the Takes window (**Takes…** in the node's right-click menu). 2. Preview the take you want to remove. 3. Open **More take actions** and choose **Delete Take** with its number. 4. Read the check, then confirm. Nodespell first checks what uses the take: - If nothing uses it, it is removed from the history for everyone in the project. - If the node currently uses it, you choose what the node uses instead before deleting. - If it is the node's last take, you confirm that the node will have no takes left; the next run must generate a new one. - If a pinned branch, a saved snapshot, a copy or publication, or a running generation refers to it, the take is protected and cannot be deleted until that changes. > **Caution:** Deleting a take does not refund credits or free storage, and it cannot be undone from the Takes window. ## Read a take's details In the Takes window, each take shows its status (**Generating…**, **Failed**, **Cancelled**, **In use**) and a **Details** section with its charge status: **charged** with the number of credits, **refunded**, **charge pending** or **not charged**. A take made with settings that differ from the node's current ones is marked **older settings**. For nodes with several outputs, choose the output from the **Output** menu. ## Frequently asked questions ### Does previewing a take change my graph? No. Previewing only changes what you see. The graph changes only when you click **Use**, **Keep this take** or **Follow latest**. ### Why can't I use a take? The take may still be generating or loading, it may have failed, or you may be viewing the project rather than editing it. Results made before take numbers existed can be previewed but cannot be used as a whole take. ### Will a new run replace the take I chose? Not if the node is on a kept take. New takes are added to the history and the node stays on the take you chose until you choose **Follow latest**. ### Does deleting a take give my credits back? No. Deleting removes the take from history but does not refund credits or free storage. ### Why does my older project not show takes? Takes are turned on per project, and new projects start with them. Projects without takes keep their earlier result history, and you keep a result there with **Lock Result** under **Flags** in the node's right-click menu. ## Related - [Run nodes](https://docs.nodespell.com/graph/run-nodes/): what a run reuses and what it generates - [Branch and switch presentation](https://docs.nodespell.com/graph/branch-and-switch-presentation/): change model without losing takes - [Viewers](https://docs.nodespell.com/edit/viewers/): inspect results in the Design panel - [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/): why snapshots protect takes - [Glossary](https://docs.nodespell.com/reference/glossary/): take, run and related terms --- # Change model or presentation without losing work > Switch a Nodespell node between a Creative Tool and its model-node view, change its model safely, and branch so earlier takes and downstream work stay intact. Source: https://docs.nodespell.com/graph/branch-and-switch-presentation/ · Last updated: 2026-11-10 In Nodespell, an AI model node can be shown as a guided [Creative Tool](https://docs.nodespell.com/creative-tools/) or as its plain model node, and you can switch between the two at any time without changing the node. Changing the model is different: once a node has results, Nodespell creates a branch, a new node with the new model, so the original node, its [takes](https://docs.nodespell.com/graph/takes/) and everything downstream stay exactly as they were. ## Presentation versus model: what changes? | Action | Same node? | Takes | Connections | Settings | | ------------------------------------------------ | --------------------- | -------------------- | -------------------------------------------------------------- | ---------------------- | | Switch presentation (Creative Tool ↔ model node) | Yes | Kept | Kept | Kept | | Change model before the node has results | Yes | None yet | Compatible ones kept | Compatible ones kept | | Change model after the node has results | No, a new branch node | Stay on the original | Inputs reconnected to the branch; outputs stay on the original | Compatible ones copied | A Creative Tool changes the editing experience, not the underlying model, pricing, results or connections. ## Switch between Creative Tool and model node 1. In **Graph**, right-click an AI model node. 2. Choose **Show as Creative Tool**. If the model fits more than one Creative Tool, pick one from the submenu. 3. To go back, right-click the node and choose **Show as** followed by the model name and **node**, for example **Show as Nano Banana 2 node**. The option only appears for models that belong to at least one Creative Tool. Cinematic Shot, Variations and Relight are offered for a node only when its prompt already contains that tool's instructions. ## Change the model before the first run With a Creative Tool node selected, choose a different model from the model picker in [Parameters](https://docs.nodespell.com/graph/parameters/). If every setting and connection carries over, the change happens in place, and a notice summarizes any values that had to be adapted to the new model. If something cannot carry over, **Review what will change** opens first. It lists **Settings that won't carry over** and **Connections that won't carry over**, with each connection marked **Incoming** or **Outgoing**. Click **Change model** to apply, or **Cancel** to keep the current model and everything you entered. ## Create a branch when the node already has results When the node already has run history, changing its model opens **Create a new model branch?**. 1. Choose the new model in the Creative Tool's model picker. 2. Read the dialog. It says how many settings and connections will not carry over to the branch, and confirms that the original node, its selected result and its downstream graph stay unchanged. 3. Click **Create branch**. Nodespell adds the branch to the right of the original. It has the same name and Creative Tool, the settings that fit the new model, and the same incoming connections. Its outputs start unconnected, so connect the branch to downstream nodes when you are ready to use it. The whole branch is one step in undo history. If the node changes while the dialog is open, Nodespell shows **The node changed while this dialog was open. Review the latest details, then confirm again.** > **Tip:** To compare models on the same prompt, create a branch for each model, run them, and view the results side by side. See [Compare models side by side](https://docs.nodespell.com/recipes/compare-models-side-by-side/). ## Frequently asked questions ### Does switching to a Creative Tool lose my results? No. Switching presentation keeps the same node, so its takes, connections, settings and price stay the same. ### Why did changing the model create a new node? The node already had run history. Nodespell branches so the earlier takes stay attached to the model that made them and the downstream graph keeps working. ### What happens to downstream connections when I branch? They stay on the original node. The branch receives the same inputs, and you connect its outputs yourself. ### Can I undo a branch? Yes. Creating the branch is a single undo step. ### Why is Show as Creative Tool missing from the menu? The node's model is not part of any Creative Tool, or the node is not an AI model node. ## Related - [Creative Tools](https://docs.nodespell.com/creative-tools/): every tool and how it relates to nodes - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/): the model picker - [Creative Tools on the canvas](https://docs.nodespell.com/creative-tools/on-the-canvas/): working with tool nodes in Graph - [Takes](https://docs.nodespell.com/graph/takes/): results stay with the original node - [Revise without starting over](https://docs.nodespell.com/recipes/revise-without-starting-over/): a worked example --- # Track runs, failures and credit use in Tasks > Use the Nodespell Task queue and Tasks panel to follow runs in progress, stop or resume them, read failure details, and review credit usage. Source: https://docs.nodespell.com/graph/tasks/ · Last updated: 2026-11-10 Nodespell tracks generation work in two places. The Task queue in the graph toolbar shows runs that are in progress right now and lets you stop or resume them. The Tasks panel is a history of your AI tasks, with status filters, failure details, and the credits your generations and [Agent](https://docs.nodespell.com/agent/) turns have used. ## See what is running now When a run is in progress, a **running** indicator with a count appears in the graph toolbar, for example **2 running**. Click it to open **Task queue**. Each run in the queue shows: - the run's status and how long it has been going; - a progress bar and the number of finished steps, for example `3/5`; - up to three of the steps it is working on next, each with the node name and its status; - **Stop** to cancel the run, and **Resume** when the run is waiting for this browser. | Run status | What it means | | ---------------------------------------------------- | ------------------------------------------------------------------------------------------ | | **Queued** | Accepted and waiting to start. | | **Planning** / **Planning next steps** | Working out the steps to run. | | **Running** | Steps are executing. | | **Waiting** | Waiting for an earlier step before continuing. | | **Waiting for browser** / **Resume browser session** | Part of the work runs in your browser tab. Click **Resume** to continue it in this window. | | **Waiting for confirmation** | Waiting for you to confirm before it continues. | Some steps run in your browser tab rather than on the server, so keep the tab open until they finish. AI generations run on the server and keep going if you close the tab. ## Stop a run 1. Click the **running** indicator in the graph toolbar. 2. Find the run in **Task queue**. 3. Click **Stop**. A stopped generation appears in the node's [takes](https://docs.nodespell.com/graph/takes/) as **Cancelled**, and its details show whether it was charged. ## Open the Tasks panel Open a new panel from any panel's **New panel** menu and choose **Tasks** under **Automation**. The panel has three tabs: | Tab | What it shows | | ------------ | --------------------------------------------------------------------------------------------------------------------------------------------------- | | **Tasks** | Your AI tasks, newest first, 20 per page, with **Previous** and **Next**. | | **Usage** | **Credit Usage**: individual credit deductions for AI tasks and Agent turns. **Load older** and **Newer** page through the latest 1,000 activities. | | **Timeline** | **Credit usage over time**, with a **Time range** from **Last 4 hours** to **Last 10 years**. | The panel refreshes on its own when your connection recovers. Use the refresh button to update it immediately. ## Find a task and read its details In the **Tasks** tab, use the search box, the model type filter (**Text Models**, **Image Models**, **Video Models**, **Audio Models**) and the status buttons (**All Statuses**, **Completed**, **Running**, **Queued**, **Failed**, **Pending**) to narrow the list. Each row shows the model, the status, how long ago it ran, how long it took, and the last characters of its task ID. Click a row to show its details: | Detail | Description | | ----------------------------- | ---------------------------------------------------------------------------------- | | **Model** | The model that ran. | | **Task ID** | The full ID, with **Copy Task ID**. Include it when you contact support. | | **Start Time** / **End Time** | When the task started and finished. | | **Status hint** | Extra status information, when there is any. | | **Error** | For a failed task, a readable message, the error code, and **Copy Error Message**. | ## What to do when a task fails 1. Read the error in the Tasks panel, or right-click the node and choose **Tasks** › **Copy error**. 2. Fix the cause. Common causes are missing or unfinished input uploads, an input of the wrong media type, a prompt the model does not accept, or not enough credits. 3. Run the node again. There is no separate retry button; a new run creates a new take, and the failed one stays in history marked **Failed**. See [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/) for specific messages and their fixes. > **Note:** The Tasks panel lists the AI tasks for your account. Free work such as compositing does not appear there; check the node itself for those results. ## Frequently asked questions ### Where do I see generations that are still running? Click the **running** indicator in the graph toolbar to open **Task queue**. It lists every run in progress with its steps and progress. ### How do I retry a failed generation? Fix the cause shown in the error, then run the node again. Each run creates a new take, so the failed attempt stays in the history. ### Does closing the browser stop my generations? AI generations continue on the server. Steps that run in the browser pause until you reopen the project and click **Resume** in **Task queue**. ### Where can I see how many credits I used? Open the Tasks panel and choose **Usage** for individual charges or **Timeline** for usage over time. ### What should I send support about a failed task? Copy the task ID with **Copy Task ID** and the message with **Copy Error Message** from the task's details, and include both. ## Related - [Run nodes](https://docs.nodespell.com/graph/run-nodes/): start, plan and batch runs - [Takes](https://docs.nodespell.com/graph/takes/): failed and cancelled takes in node history - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): how credits are charged - [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/): fixes for common failures - [Support](https://docs.nodespell.com/account/support/): what to include when you contact support --- # Work with text, numbers and logic nodes > A guided map of Nodespell's data and logic nodes for text, numbers, arrays, objects and JSON, math, conditions and loops, with worked examples. Source: https://docs.nodespell.com/graph/data-and-logic/ · Last updated: 2026-11-10 Data and logic nodes in Nodespell work with values instead of media: text, numbers, true/false values, lists (arrays) and key-value objects. Use them to build prompts from parts, calculate parameter values, choose between inputs, and repeat work for every item in a list. They run in your browser and never use [credits](https://docs.nodespell.com/account/plans-and-credits/). Add any of them with the [Tab menu](https://docs.nodespell.com/graph/add-nodes/) by typing the label shown below, or browse them in the [Node Browser](https://docs.nodespell.com/graph/add-nodes/). The output of a data node connects to any compatible input or [parameter](https://docs.nodespell.com/graph/parameters/), including a model's prompt. ## Text nodes | Node | Use it to | | ------------------------------------------------------------ | -------------------------------------------------------------------------------------------- | | **String Template** | Fill a template such as `A photo of ${0} at ${1}` with the strings connected to **Strings**. | | **String Concat** | Join several strings, with an optional **Prefix**, **Separator** and **Suffix**. | | **String Replace** | Replace text; wrap the search in `/…/` to use a regular expression. | | **String Split** | Split text into an array at a **Separator**. | | **String Case** | Change case, for example **Upper Case**, **Sentence Case** or **Kebab Case**. | | **String Trim**, **String Pad**, **String Slice** | Clean up, pad or cut text. | | **String Length**, **String Includes**, **String Ends With** | Measure or test text. | ## Number and math nodes | Node | Use it to | | ----------------------------------------------------------------------------- | -------------------------------------------------------------------- | | **Plus**, **Minus**, **Multiply**, **Divide**, **Remainder**, **Exponent** | Basic arithmetic on **First** and **Second**. | | **Remap** | Map a value from one range to another, for example 0–1 to 0–100. | | **Clamp** | Keep a value between **Min** and **Max**. | | **Round**, **Floor**, **Ceiling**, **Truncate** | Turn a number into a whole number. | | **Random** | Return a number between 0 and 1; set **Seed** to make it repeatable. | | **Noise** | Return a smooth noise value between 0 and 1. | | **Minimum**, **Maximum**, **Absolute**, **Negate**, **Sign**, **Square Root** | Other common operations. | | **Is Finite**, **Is Integer**, **Is NaN** | Test a number. | ## List (array) nodes | Node | Use it to | | ------------------------------------------------------------------------------------- | ------------------------------------------------------------ | | **Merge As Array** | Collect several values into one array. | | **Array Get Element** | Read one item by index (0 is the first). | | **Array Length** | Count the items. | | **Array Range** | Make a sequence of numbers, such as 0 to 10 in steps of 0.5. | | **Array Slice**, **Array Take** | Keep part of an array. | | **Array Filter**, **Array Deduplicate** | Remove empty, unwanted or repeated items. | | **Array Sort**, **Array Reverse**, **Array Shuffle** | Reorder items; **Array Shuffle** accepts a seed. | | **Array Append**, **Array Insert Item**, **Array Remove Item**, **Array Set Element** | Edit items. | | **Array Concat**, **Array Flatten**, **Array Chunk**, **Array Zip**, **Repeat** | Combine, flatten, split or repeat arrays. | ## Object and JSON nodes | Node | Use it to | | ----------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------- | | **JSON Parse** | Turn JSON text, such as a text model's structured answer, into an object. | | **JSON Stringify** | Turn an object back into text. | | **Object Get Key** | Read one value by its key. | | **Create Object**, **Key Value Pair** | Build an object. | | **Object Add Key**, **Object Delete Key**, **Object Rename Key**, **Object Merge**, **Object Filter** | Edit an object. | | **Object Entries**, **Object From Entries**, **Array Zip Object** | Convert between objects and arrays. | ## Logic, flow control and conversion nodes | Node | Use it to | | --------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------ | | **If Else** | Output **True Value** or **False Value** depending on **Condition**. | | **Compare** | Compare two numbers with **Equal**, **Less Than**, **Greater Than** and similar operations. | | **String Compare** | Compare two strings, optionally ignoring case or whitespace. | | **Gate** | Combine two true/false values with **AND**, **OR**, **XOR**, **NOT** and others. | | **Switch Input** | Pass through one of several connected inputs, chosen by its index in **Which** (0 is the first). | | **Is Falsy**, **Is Nullish** | Test whether a value is empty or missing. | | **Map Container** | Run the nodes inside it once for every item in an array. | | **String To Float**, **String To Int**, **To String**, **To Fixed**, **To Precision**, **To Exponential** | Convert between text and numbers. | | **Constant** | Hold a fixed **String**, **Number** or **Boolean** value. | | **Reference** | Read another node's output without a connection. | | **Type Of** | Report the type of a value. | | **Dot**, **No Operation** | Pass a value through unchanged to tidy connections. | ## Example: build a prompt from a template 1. Add two **Constant** nodes with **Value Type** set to **String**, for example `a red bicycle` and `golden hour`. 2. Add a **String Template** node and set **Template String** to `Product photo of ${0}, lit at ${1}`. 3. Connect both constants to its **Strings** input, in order. The first connection fills `${0}`, the second `${1}`. 4. Connect the template's **Value** output to the prompt input of an image model and run it. Change either constant and run again to try a new subject or light without retyping the prompt. For writing longer prompts by hand, see [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/). ## Example: repeat work for every item in a list 1. Add a **Constant** with the text `cat, dog, fox` and connect it to a **String Split** node with **Separator** set to `, `. 2. Add a **Map Container** and connect the split **Output** to its **Items** input. 3. Select the Map Container and press `Enter` to go inside. 4. Add a **Map Item** node (the current item), connect it to a **String Case** node set to **Upper Case**, and connect that to a **Map Output** node. 5. Press `U` to go back out, select the Map Container, and run it. The **Values** output returns the results in the original order: `CAT`, `DOG`, `FOX`. Inside the container, **Map Index** gives the item's position and **Map Length** the total. **Max Concurrency** (default 4, up to 16) sets how many items run at once. Right-click the container and choose **Inspect Map Runs** to see each item's result. > **Note:** Run the Map Container itself. Running a node inside it on its own shows **Run Map body node directly?**, because a direct run does not update the Map's output. ## Example: choose between two values 1. Connect a number, such as a **Random** node with a seed, to **Left** on a **Compare** node. Set **Right** to `0.5` and **Operation** to **Greater Than**. 2. Connect the **Compare** output to **Condition** on an **If Else** node. 3. Connect two prompts, one to **True Value** and one to **False Value**, and use the **If Else** output downstream. For a fixed choice from several options, use **Switch Input** and set **Which** instead. ## Example: read a field from JSON Connect a text model's output that contains JSON to **JSON Parse**, then connect the parsed object to **Object Get Key** and set **Key**, for example `title`. The output is that field's value, ready to use in a **String Template** or as a parameter. ## Frequently asked questions ### Do data and logic nodes cost credits? No. They calculate in your browser and appear in run plans as free updates. ### How do I insert values into a prompt? Use **String Template** with `${0}`, `${1}` and so on in **Template String**, and connect the values to **Strings** in order. You can also write [expressions](https://docs.nodespell.com/graph/expressions/) directly in parameters. ### How do I loop over a list in Nodespell? Put the work inside a **Map Container**, use **Map Item** for the current item and **Map Output** for each result, and connect your array to **Items**. ### What is the difference between Constant and a typed parameter? A **Constant** is a node whose output you can connect to many inputs, so one value can drive several nodes. A parameter value belongs to a single node. ## Related - [Expressions](https://docs.nodespell.com/graph/expressions/): calculate parameter values inline - [Connections](https://docs.nodespell.com/graph/connections/): which outputs fit which inputs - [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/): going inside nodes with `Enter` and `U` - [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/): the Prompt node --- # Save workflows, presets and node defaults > Reuse work in Nodespell by saving selected nodes as a Workflow, applying presets to nodes, and setting your own default settings for new nodes. Source: https://docs.nodespell.com/graph/snippets-and-presets/ · Last updated: 2026-11-10 Nodespell gives you three ways to reuse work in the graph. A Workflow is a saved group of nodes and connections you can add to any project. A preset is a named setup for a single node that you apply on purpose. A node default is the set of values every new node of that type starts with. ## Which one should I use? | You want to… | Use | Where | | ----------------------------------------------- | -------------- | ---------------------------------------------------- | | Reuse several connected nodes in other projects | A Workflow | **Workflows** menu › **Save As Workflow** | | Apply a named look or setup to a node | A preset | Tab menu, Node Browser, or **Presets** in Parameters | | Make every new node of a type start your way | A node default | Node right-click menu › **Save as my default…** | ## Save selected nodes as a Workflow 1. In **Graph**, select the nodes you want to keep. 2. Open the **Workflows** menu in the graph toolbar and choose **Save As Workflow**. You can also right-click a node and choose **Advanced** › **Presets** › **Create**. 3. In **Create New Workflow**, enter a **Name** and, optionally, a **Category**, **Tags**, **Keywords** and a **Description** (Markdown is supported). 4. Optionally add a **Cover** and **Thumbnails** from your image Assets. 5. Click **Create**. Nodespell saves the nodes, their settings and the connections between them. A Workflow saved from exactly one node is also a preset for that node type. ## Add a saved Workflow to a project 1. Open the **Workflows** menu and choose **Workflow** to open the Workflow browser. 2. Search, or use **Filters**: **Official Only**, **Favorites Only**, **Presets Only**, **Tags** and **Keywords**. Sort by **Relevance**, **Name**, **Category**, **Date Created** or **Update Time**. 3. Click **Use** on a card, or right-click it and choose **Use**, to add a copy of its nodes to the current graph. Each card shows the node and connection count. Click **Favourite** on a card to find it again with **Favorites Only**. Official Workflows are marked **Official**. ## Apply a preset to a node Presets set a node's parameters to a saved setup. - **From your Workflows**: select one or more nodes of the same type, open the **More** menu in [Parameters](https://docs.nodespell.com/graph/parameters/), and choose a preset under **Presets**. Presets you saved and official presets for that node type are listed. In the Workflow browser, a preset card also offers **Apply to Selected Nodes** when the selection matches its node type. - **Built-in presets**: in the [Tab menu](https://docs.nodespell.com/graph/add-nodes/) and the Node Browser, nodes with built-in presets show them beneath the node, tagged **Preset**. Choosing one adds the node with that setup. | Built-in preset | Node | | ---------------------- | ------------------------- | | **Warm Contrast** | Color Correct | | **Soft Bloom** | Glow | | **Soft Halo** | Exponential Blur | | **Speed Streak** | Directional Blur | | **Optical Flow** | Motion Blur | | **Lens Soften** | Defocus | | **Lower Third** | Text | | **Animated Grain** | Noise | | **Vignette Backplate** | Gradient | | **Half Speed** | Retime | | **RG Displacement** | STMap | | **Handheld** | Camera Shake | | **Folder** | Level, set up as a Folder | ## Save your own default settings for a node type 1. Set up a node the way you like it. 2. Right-click the node and choose **Save as my default…** (or **Update my default…** if you already have one). 3. In the dialog, tick the settings to save. You can also save **Appearance** such as color and node size, and the **On-node glance** values shown on the node. Settings you leave unticked keep the Nodespell default. 4. Click **Save default**. Your default applies to nodes of that type you add from now on, in any project. Existing nodes do not change. For a Creative Tool node, the default can also set the default model. Some settings cannot be saved, and the dialog lists why under **Can't be saved**: a value that is connected to another node, uses an [expression](https://docs.nodespell.com/graph/expressions/), references a project file or other nodes, or is animated with keyframes. To review, turn off or remove defaults, open **Preferences** (`Cmd/Ctrl+,`) and choose **My node defaults**. **Revert Parameters to Default** in the Parameters **More** menu restores your saved defaults; settings you have not saved go back to the Nodespell default. ## Frequently asked questions ### What is the difference between a preset and a node default? A preset is a named setup you apply on purpose. A node default is applied automatically to every new node of that type you add. ### Do node defaults change nodes I already have? No. A saved default only affects nodes you add after saving it. ### Who can edit or delete a Workflow? You can **Edit** and **Delete** Workflows you created. Official Workflows can be used and favourited but not changed. ### Why is Save as my default unavailable? Select a single node. Defaults are saved from one node at a time, and some node types do not support defaults. ## Related - [Parameters](https://docs.nodespell.com/graph/parameters/): editing, resetting and pinning values - [Add nodes](https://docs.nodespell.com/graph/add-nodes/): the Tab menu and Node Browser - [Templates and sharing](https://docs.nodespell.com/projects/templates-and-sharing/): publishing work for others - [Settings](https://docs.nodespell.com/account/settings/): Preferences, including My node defaults - [Organize your graph](https://docs.nodespell.com/graph/organize/): Groups and layout --- # Composite images and video with compositors > Build layered composites in Nodespell with Image Compositor and Compositor nodes, Merge blend modes, custom channel expressions and canvas sizing. Source: https://docs.nodespell.com/edit/compositing/ · Last updated: 2026-11-10 A compositor in Nodespell is a node that holds its own node graph for layering images and video, much like a Nuke script inside one node. **Image Compositor** produces a still image and **Compositor** produces a video. Everything inside renders on the GPU in your browser, and the result leaves through one output that the rest of your graph, including AI generators, can use. ## Which compositor should you use? Nodespell has two compositor nodes. Both hold a node graph, both use the same effect nodes, and you can convert one into the other. | | **Image Compositor** | **Compositor** | | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------- | -------------------------------------------------------------------------- | | Produces | A still image (output **Image**) | A video (outputs **Video**, **Compositor** and **Preview Frame**) | | Input on the node | **Images** | **Assets** | | Timeline, playhead and keyframes | No | Yes, with **Frame Rate** (default 30) and **Duration Seconds** (default 5) | | Time-based nodes (Retime, Time Offset, Freeze Frame, Reverse Time, Optical Flow, Motion Blur, Camera Shake, Switch/Dissolve, Image Sequence, Timeline Source, Compositor Source) | Not supported | Supported | | Audio | None | **Audio Source**: **First Input** or **None** | | **Result** choices with colour management on | **Review PNG** or **Master EXR** | **Review video**, or **HDR review** where your device supports it | If an Image Compositor contains a time-based node, it cannot render a still and names the node so you can remove it or convert back to Compositor. The menus that convert between the two call the video compositor **Video Compositor**; it is the same node as **Compositor**. ## Create a compositor 1. In the Graph, press `Tab` to open the [Tab menu](https://docs.nodespell.com/graph/add-nodes/) and add **Image Compositor** or **Compositor**. 2. Connect your images or videos to the compositor's input (**Images** or **Assets**). The input accepts several connections, in order. 3. Select the compositor and press `Enter` to go inside it. Press `U` to come back out. 4. Inside, a new compositor already contains a **Compositor Input** connected to a **Compositor Output**. Add effect and Merge nodes between them. 5. Select the compositor or any node inside it to see the composite in the [Design panel](https://docs.nodespell.com/edit/viewers/). You can also build a chain of image effects in the Graph first, select those nodes, right-click and choose **Convert to Video Compositor** or **Convert to Image Compositor**. Nodespell moves the selected nodes into a new compositor. If some connections cannot be carried inside, a message says how many were removed; use Undo to restore them. To switch an existing compositor between still and video, use the **Actions** submenu on the compositor node. ## How do Compositor Input and Compositor Output work? **Compositor Input** brings the media connected to the parent compositor into its inner graph. Its **Index** parameter is zero-based: index 0 is the first connection on the compositor's input, index 1 the second, and so on. It has two outputs, **Asset** (the one item at that index) and **Assets** (all of them). Add one Compositor Input per source you want to work with separately. **Compositor Output** marks what the compositor delivers. Whatever you connect to its **Source** input becomes the compositor's result. On a video Compositor, **Use In/Out Points** (on by default) renders the playback range; turn it off to set **Start Frame** and **End Frame** yourself. A video Compositor also offers source nodes for other kinds of media: **Image Set Source** (several stills for switching or dissolving), **Image Sequence** (stills spread over time by **Total Length**, **Each Image Length** or **Manual Index**), **Timeline Source**, **Compositor Source** (a nested compositor) and **Switch/Dissolve**. ## Layer images with Merge **Merge** combines one **Background** with one or more **Foregrounds**. Each foreground is merged over the result so far, in connection order, using the chosen **Mode** (default **Over**). **Mix** (0–1, default 1) fades the merged result back toward the background. Merge follows Nuke's conventions: A is the foreground and B is the background. Foregrounds are premultiplied before merging, so you can feed straight-alpha images directly; only **Matte** reads the foreground unpremultiplied. Inputs whose size differs from the canvas are scaled to fit inside it without cropping. Outside a compositor, the **Composite** node in the Graph does the same job for still images, with the same modes and matching results, including transparent inputs. Composite also has **Edit Composite**, where you move, rotate and scale each foreground layer. ## Blend modes in Merge and Composite In the formulas, A and B are the foreground and background colour values per channel, and a and b are their alphas. Values are not clamped, so float and HDR values pass through. | Mode | Result | Typical use | | --------------------- | ------------------------------------------------------ | -------------------------------------------- | | **Over** | A + B × (1 − a) | Standard layering of a premultiplied element | | **Under** | A × (1 − b) + B | Put A behind B | | **Atop** | A × b + B × (1 − a) | Layer A only where B exists | | **In** | A × b | Keep A inside B's alpha | | **Out** | A × (1 − b) | Keep A outside B's alpha | | **Mask** | B × a | Keep B inside A's alpha | | **Stencil** | B × (1 − a) | Cut A's shape out of B | | **XOR** | A × (1 − b) + B × (1 − a) | Show each only where the other is absent | | **Conjoint Over** | Over that treats overlapping alphas as shared coverage | Layers cut from the same matte | | **Disjoint Over** | Over that treats alphas as non-overlapping | Layers that meet edge to edge | | **Matte** | Mix B toward unpremultiplied A by a | Straight-alpha foregrounds | | **Copy** | A | Replace with the foreground | | **Add** / **Plus** | A + B | Light, glows, fire, flares | | **Minus** | A − B | Subtract the background from A | | **From** | B − A | Subtract A from the background | | **Multiply** | A × B | Shadows, darkening, applying dirt | | **Divide** | A ÷ B | Removing a multiplied pass | | **Screen** | A + B − A × B (Max when either is outside 0–1) | Soft brightening | | **Overlay** | Multiply or screen, chosen by B | Contrast and texture | | **Hard Light** | Multiply or screen, chosen by A | Strong texture overlays | | **Soft Light** | W3C soft-light formula | Gentle contrast | | **Color Dodge** | Brighten B by A | Hot highlights | | **Color Burn** | Darken B by A | Deep, saturated darks | | **Difference** | \|A − B\| | Checking alignment or changes | | **Exclusion** | A + B − 2 × A × B | Low-contrast difference look | | **Average** | (A + B) ÷ 2 | Blending two plates equally | | **Geometric** | 2 × A × B ÷ (A + B) | Averaging that favours the darker value | | **Hypot** | √(A² + B²) | Combining reflections | | **Max** / **Lighten** | Larger of A and B | Keep the brightest | | **Min** / **Darken** | Smaller of A and B | Keep the darkest | | **Custom** | Your own expressions | Per-channel control | ## Write a custom merge with channel expressions Set **Mode** to **Custom** to combine inputs with your own maths. Two fields appear: **Colour** (default `A.rgb`) and **Alpha** (default `A.a`). > **Caution:** In Custom mode the letters change meaning. **A is the background**, and B, C, D and so on are the foregrounds in connection order, up to H. This differs from the blend-mode formulas above. - **Pick channels** with `.rgb`, `.r`, `.g`, `.b`, `.a`, or combinations such as `.rg`. - **Colour** must give three numbers, or one number for grey. **Alpha** must give one number. - **Position and time:** `x` and `y` (pixels, y counted up from the bottom), `u` and `v` (0 to 1), `width`, `height`, `frame` and `time` (seconds). - **Operators:** `+ - * / ^`, comparisons, `&&`, `||`, `!` and `condition ? a : b`. - **Functions** include `abs`, `floor`, `ceil`, `round`, `fract`, `sqrt`, `exp`, `ln`, `log2`, `sin`, `cos`, `tan`, `atan2`, `pow`, `min`, `max`, `mod`, `step`, `clamp`, `saturate`, `lerp`, `smoothstep`, `fit`, `luma`, `dot`, `length`, `premult`, `unpremult`, `vec2` and `vec3`. Inputs in Custom mode are straight (unpremultiplied) alpha, and so is the result. An input the expression reads but nothing is connected to reads as transparent. **Mix** sets how much of the custom result replaces input A. Expressions are limited to 600 characters. | Goal | Colour | Alpha | | ------------------------------------------------- | ------------------------- | --------------- | | Blend B over A using C's red channel as the matte | `lerp(A.rgb, B.rgb, C.r)` | `max(A.a, B.a)` | | Take red from B, keep A's green and blue | `vec3(B.r, A.g, A.b)` | `A.a` | | Left half A, right half B | `u < 0.5 ? A.rgb : B.rgb` | `A.a` | ## Set the canvas size **Canvas size** sets the compositor's frame. New compositors start on **First input**; compositors saved before canvas sizing existed stay on **Custom**. | Option | Canvas | | -------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **First input** | The dimensions of the first connected input | | **Largest input** / **Smallest input** | The input with the most or fewest pixels; ties keep connection order | | **Custom** | Your own width and height (1920 × 1080 by default). The first switch to Custom starts from the current size | | Presets | 720p Landscape · 1280 × 720, 1080p Landscape · 1920 × 1080, 4K UHD Landscape · 3840 × 2160, 1080p Portrait · 1080 × 1920, Square · 1080 × 1080, DCI 4K · 4096 × 2160 | Automatic sizes use the original media dimensions, including orientation, never a thumbnail's size. While dimensions are loading, the compositor shows **Waiting for input dimensions.** You can also change the size from the compositor's right-click menu under **Canvas size**. Changing the canvas size does not change frame rate, duration or colour settings. ## How does a composite feed generators downstream? Connect the compositor's output to any node that accepts an image or video. An Image Compositor's **Image** output carries the still; a video Compositor's **Video** output carries the rendered clip and **Preview Frame** carries a still of the current frame. When you edit a composite, downstream generators receive the updated result the next time they run. Compositing runs in your browser, not on Nodespell's servers. What the compositor delivers depends on its colour settings; see [Color management](https://docs.nodespell.com/edit/color-management/). ## Frequently asked questions ### Does compositing use credits? Compositor rendering and its effect nodes run on your device's GPU in the browser. Credits apply to AI model nodes, such as a generator you feed the composite into, not to compositing itself. ### Why can't I find some nodes in the Tab menu inside a compositor? Inside a compositor, the Tab menu shows only nodes that work in compositors. Inside an Image Compositor it also hides time-based nodes such as Retime and Camera Shake, which need a video Compositor. ### Can I animate a Merge or effect inside an Image Compositor? No. An Image Compositor has no timeline. Convert it to a video Compositor to set [keyframes](https://docs.nodespell.com/edit/keyframes/). ### What is the difference between Merge and Composite? Merge works inside compositors; Composite works in the ordinary Graph on still images. They share the same blend modes and give matching results. ### Why does the Custom mode use A for the background? Custom mode names inputs in connection order, starting with the Background as A. The fixed blend modes use Nuke's naming, where A is the foreground. ## Related - [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/): build mattes for Merge and effects. - [Transforms and crop](https://docs.nodespell.com/edit/transforms-and-crop/): position and size layers. - [Color management](https://docs.nodespell.com/edit/color-management/): working space, viewing and delivery. - [Keyframes](https://docs.nodespell.com/edit/keyframes/): animate compositor parameters. - [Viewers](https://docs.nodespell.com/edit/viewers/): inspect the composite in the Design panel. - [Composite and finish recipe](https://docs.nodespell.com/recipes/composite-and-finish/): a worked example. --- # Create mattes, key and mask effects > Make and refine mattes in Nodespell with Keyer, Erode/Dilate, Roto, Apply Mask and Mask Mix, and limit any effect with its Mask input. Source: https://docs.nodespell.com/edit/masks-and-keying/ · Last updated: 2026-11-10 A matte in Nodespell is a grayscale image or alpha channel that says where something applies: white is fully on, black is off, gray is partial. Nodespell builds mattes with **Keyer**, **Roto** and AI cutouts, refines them with **Erode/Dilate**, and uses them to cut out images, mix two images, or limit any effect through its **Mask** input. ## Alpha, straight and premultiplied Every image node in Nodespell carries RGBA. The alpha channel is the image's own matte. Two conventions matter when you combine layers: - **Straight (unpremultiplied)** alpha keeps full colour values in RGB, with alpha stored separately. - **Premultiplied** alpha stores RGB already multiplied by alpha, which is what Over-style merges expect. Merge and Composite premultiply each foreground for you, and Custom merge expressions see straight alpha, so most graphs need no manual step. When you do need to switch, use **Premultiply** (multiplies RGB by alpha) and **Unpremultiply** (divides RGB by alpha). To move a matte between images, **Channel Copy** writes any source channel, or its luminance, into the target's R, G, B or alpha, much like Nuke's Copy or Shuffle. ## Pull a key with Keyer **Keyer** writes a new matte from the image itself. 1. Add **Keyer** after your plate and connect the image. 2. Choose a **Mode** (table below). For **Chroma Distance**, set **Screen Color** to the backing colour. 3. Set **Output** to **Matte View** to see the matte as black and white. 4. Raise **Clip Black** until the background is solid black, and lower **Clip White** until the subject is solid white. **Softness** widens the transition between them. 5. Turn on **Invert** if the matte is the wrong way round. 6. Set **Output** back to **Write Alpha** or **Premultiplied RGB**. | **Mode** | Matte is based on | Use for | | ------------------------------------------------- | ---------------------------------------------------------------------- | ------------------------------------------------------ | | **Luminance** (default) | Brightness of each pixel | Bright elements on dark (fire, sparks, sky) | | **Chroma Distance** | Distance from **Screen Color** | Any solid-colour backing | | **Green Screen**, **Blue Screen**, **Red Screen** | How much that channel exceeds the other two; the backing goes to black | Coloured screens | | **Saturation** | Colour saturation | Separating colourful subjects from neutral backgrounds | | Control | Default | What it does | | ----------------- | --------------- | --------------------------------------------------------------------------------------------------------------------------------- | | **Clip Black** | 0.08 | Matte values at or below this become 0 | | **Clip White** | 0.35 | Matte values at or above this become 1 | | **Softness** | 0.02 | Extends the ramp between the clip points (0–0.5) | | **Combine Alpha** | **Replace** | **Replace** discards the input alpha; **Intersect** multiplies the new matte with it; **Union** combines their coverage | | **Output** | **Write Alpha** | **Write Alpha** keeps RGB and writes alpha; **Matte View** shows the matte; **Premultiplied RGB** multiplies RGB by the new alpha | | **Mix** | 1 | Blends between the original alpha and the new matte | Keyer does not suppress spill. Reduce green or blue fringing afterwards with [Hue Correct](https://docs.nodespell.com/edit/color-correction/) or a grade limited by the matte. ## Grow or shrink a matte with Erode/Dilate **Erode/Dilate** shrinks or grows edges, like Nuke's Erode filter. | Control | Options | Notes | | ------------- | -------------------------------------------- | ---------------------------------- | | **Operation** | **Dilate** (default), **Erode** | Dilate grows, Erode shrinks | | **Channel** | **Alpha** (default), **Luminance**, **RGBA** | Which values the operation follows | | **Radius** | 0–16, default 2 | Measured in whole pixels | | **Mix** | 0–1 | Blend with the original | Erode a key by one or two pixels to remove a dark fringe, or dilate a garbage matte so it safely covers the subject. ## Draw shapes with Roto **Roto** draws Bezier and straight-line shapes as a matte. It is available inside compositors only. - Select the Roto node and draw in the compositor viewer with the **Pen**, **Rectangle** or **Ellipse** tools. With the Pen, click to place points and drag while clicking to draw a curve; click the first point or press `Enter` to close the shape, or `Esc` to cancel. - Drag a point to move it, double-click a point to switch it between **Smooth** and **Sharp**, and double-click an edge to add a point. Hold `Shift` while drawing a rectangle or ellipse for a square or circle. - In the shapes panel, set each shape's **Combine** to **Add**, **Subtract** or **Intersect**, rename, lock, hide or reorder shapes, and set **Edge softness** **Inside** and **Outside**, for the whole shape or selected points. - **Output** is **Alpha Matte** (default) or **RGBA**. - To animate a shape in a video Compositor, move the playhead and select **Add key**. Editing a shape on any frame adds or updates a key there. ## Cut out or mix through a matte Two nodes apply a matte directly. Both work in the Graph and inside compositors. | Node | Inputs | What it does | | -------------- | -------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Apply Mask** | **Image**, **Mask** | White keeps the image, black makes it transparent, gray is partial. **Mask source** is **Brightness** (default) or **Alpha**. Existing transparency is kept unless **Replace transparency** is on. **Strength** fades the effect. Output size matches **Image**; the mask is stretched to fit. | | **Mask Mix** | **B (base)**, **A (masked)**, **Mask** | White shows A, black shows B, gray crossfades. **Mix** at 0 returns B. Colours are alpha-weighted to avoid fringes. Output size matches **B**. | **Light Wrap** helps a keyed element sit in its plate: it bleeds light from the image on its background input into the foreground's edges, with **Size**, **Intensity**, **Background Blur** and a **Blend** of **Screen**, **Add**, **Over** or **Soft Light**. ## Limit any effect with its Mask input Most image effect nodes, such as Color Correct, Blur, Grade and Denoise, have an optional **Mask** input and a compact **Mask** control group in Parameters. | Control | Default | What it does | | --------------- | --------- | ----------------------------------------------------------------------------------------------------------- | | **Mask Source** | **Alpha** | Which channel of the mask image drives the effect: **Alpha**, **Luminance**, **Red**, **Green** or **Blue** | | **Invert Mask** | Off | Applies the effect outside the mask instead | | **Mask Mix** | 1 | Blends between the unmasked and masked result | Many of the same nodes have **Channels**, which chooses the RGBA channels the node writes. Unselected channels pass through unchanged, so you can, for example, blur only alpha. ## Get a matte automatically For an automatic cutout of a still image, use the **Silhouette** model node, which removes the background and returns a transparent **Cutout** at the original size. Its **Edge adjustments** let you expand or contract the edge, feather it, and reduce colour fringe. Silhouette is a model node, so it runs as a task on Nodespell rather than in your browser; see [Remove background](https://docs.nodespell.com/creative-tools/remove-background/). The Cutout's alpha can feed any Mask input, Apply Mask or Mask Mix. ## Frequently asked questions ### Is there an Alpha Matte node? No separate node is named Alpha Matte. Use **Apply Mask** to cut out an image with a matte, **Mask Mix** to choose between two images, or Roto's **Alpha Matte** output for a drawn matte. ### Why is my keyed edge dark or fringed? The edge may be premultiplied twice or carrying background colour. Check that only one Premultiply happens before Merge, erode the matte by a pixel with Erode/Dilate, and reduce spill with a hue-limited correction. ### Can I key a video? Yes. Keyer works inside a video Compositor on every frame. Outside a compositor, Graph image nodes process still images. ### Which channel should a mask image use? Use **Alpha** for cutouts and Roto output, and **Luminance** for grayscale mattes such as depth maps or painted masks. ## Related - [Compositing](https://docs.nodespell.com/edit/compositing/): Merge modes and premultiplication. - [Color correction](https://docs.nodespell.com/edit/color-correction/): spill and hue-limited fixes. - [Depth and defocus](https://docs.nodespell.com/edit/depth-and-defocus/): depth maps as mattes. - [Filters and effects](https://docs.nodespell.com/edit/filters-and-effects/): which effects accept masks. - [Remove background](https://docs.nodespell.com/creative-tools/remove-background/): automatic cutouts. --- # Grade and correct color with nodes > Grade images in Nodespell with Color Correct wheels, Curves with relationship curves and a histogram, Hue Correct, Brilliance, Structure and more. Source: https://docs.nodespell.com/edit/color-correction/ · Last updated: 2026-11-10 Color correction in Nodespell is done with image nodes that change tone and colour: **Color Correct** for wheel-based grading, **Curves** and **Hue Correct** for curve-based work, and focused nodes such as **Grade**, **Exposure**, **Brilliance** and **Structure**. They run on the GPU in your browser, both in the ordinary Graph and inside [compositors](https://docs.nodespell.com/edit/compositing/). ## Which node for which job? | Job | Node | Notes | | ------------------------------------------------------- | ------------------------------ | -------------------------------------------------------- | | Balance shadows, mids and highlights with colour wheels | **Color Correct** | Lift/Gamma/Gain or Shadows/Midtones/Highlights | | Set black and white points, Nuke-style | **Grade** | Blackpoint, Whitepoint, Lift, Gain, Multiply, Add, Gamma | | Shape tone per channel | **Curves** | RGB, Red, Green, Blue and Alpha curves | | Change saturation or hue by brightness or saturation | **Curves** relationship curves | Four curves, such as Luminance → Saturation | | Change one range of hues | **Hue Correct** | Six curves driven by the original hue | | Brighten or darken in photographic stops | **Exposure** | Exposure (stops), Offset and Gamma per channel | | Classic input/output levels | **Levels** | Input black, white and gamma; output black and white | | Quick brightness and contrast | **Brightness Contrast** | Per-channel brightness and contrast | | Overall hue, saturation, lightness | **HSL** | **Hue Shift** in degrees, **Saturation**, **Lightness** | | Warm or cool the image | **Colour Temperature** | Temperature and Tint | | Open shadows and recover highlights | **Brilliance** | Positive reveals detail; negative deepens shadows | | Add or remove local "punch" | **Structure** | Local contrast that protects strong lighting edges | | Apply a LUT | **Apply LUT** | Image-strip LUT or a CUBE 1D/3D LUT, with Strength | | Darken or colour the frame edges | **Vignette** | Size, Blur, Strength and Colour | | Flip or reduce colour values | **Invert**, **Posterize** | Invert chosen channels; reduce levels | To convert between colour spaces, see [Color management](https://docs.nodespell.com/edit/color-management/). ## Grade with Color Correct **Color Correct** grades with interactive colour wheels in Parameters, with every number available under **Advanced**. Use **Open in larger window** to work on the wheels in a bigger window. | Setting | Options or default | What it does | | ---------------------------- | ---------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------- | | **Mode** | **Lift Gamma Gain** (default) or **Shadows Midtones Highlights** | Which three tonal wheels you grade | | Wheels | Red, green and blue balance plus a luma value per wheel | Balance shifts colour; luma (0–2, default 1) brightens or darkens that range | | Shadows and highlight ranges | **Shadows End** 0.045, **Shadows Softness** 0.045, **Highlights Start** 0.75, **Highlights Softness** 0.25 | Only in Shadows Midtones Highlights mode | | **Tonal Range False Colour** | Off | Shows shadows and highlights as a false-colour overlay to check the ranges | | **Saturation** | 1 (0–2) | Overall saturation | | **Contrast** and **Pivot** | 1 (0–3) and 0.18 | Contrast around a pivot value; 0.18 is scene-linear mid-grey | | **Temperature** and **Tint** | 0 (−1 to 1) | Warm/cool and green/magenta shifts | | **Mix** | 1 | Blend with the original | Lift Gamma Gain works the way colourists expect: lift moves the blacks, gamma the mids, gain the whites. Shadows/Midtones/Highlights splits the image into three ranges you set yourself. ## Shape tone with Curves **Curves** remaps channels with editable curves. The RGB curve applies first, then the individual Red, Green and Blue curves. Alpha has its own curve. All curves stay active; **Edit Curve** only chooses which one you are editing. 1. Connect an image to **Curves** and choose a curve in **Edit Curve**. 2. Double-click the graph to add a point, and drag to reshape. Input runs left to right, output bottom to top. 3. Double-click a point to remove it. Arrow keys nudge the selected point; hold `Option/Alt` while dragging to move slowly, or `Option/Alt+Shift` for very slowly. 4. Set **Interpolation** to **Smooth** or **Linear**. 5. Use **Mix** to blend the result with the original. Each curve holds up to 16 points. The end slopes continue beyond the graph, so values above 1 in float or HDR images are remapped smoothly rather than clipped. **Open in larger window** opens a bigger editor with a side list of every curve, an **Edited** marker on changed ones, **One curve** or **All curves** views, and **Show edited curves** to draw the others behind the one you are editing. ### Relationship curves Four more curves change colour by how saturated or bright each pixel already is. Each starts flat at 0.5, which means no change, and all of them read the original image, not each other's output. | Curve | Horizontal axis | Vertical axis | | --------------------------- | --------------- | ------------------------- | | **Saturation → Saturation** | Saturation | Saturation change (stops) | | **Saturation → Luminance** | Saturation | Brightness change (stops) | | **Luminance → Saturation** | Luminance | Saturation change (stops) | | **Luminance → Hue** | Luminance | Hue shift (°) | For example, pull **Luminance → Saturation** down at the right to desaturate highlights, a common fix for over-coloured skies. ### Read the histogram Turn on **Histogram** in the Curves editor to see where the input image's pixels sit behind the graph. Choose **This curve** or **R G B**, and **Linear** or **Log** scaling. The histogram shows the image going into Curves, before the remap. It is measured from a reduced-size 8-bit preview, so it always spans 0 to 1 and counts are proportional; HDR values above 1 appear clipped at the right edge. It appears once the node has an input image. ## Correct one range of hues with Hue Correct **Hue Correct** adjusts colours by their original hue, like Nuke's HueCorrect. Its six curves all use hue across the horizontal axis. | Curve | Moves colours of that hue | | ---------------------------------------------- | ---------------------------------------------------------------------------- | | **Hue → Hue** | Around the colour wheel (hue shift in degrees) | | **Hue → Saturation** | Richer (up) or paler (down) | | **Hue → Luminance** | Brighter (up) or darker (down) | | **Hue → Red**, **Hue → Green**, **Hue → Blue** | Scales that channel, to add or pull out a colour in one part of the spectrum | A curve left flat in the middle changes nothing. To place a point at a specific hue, click the spectrum, or use **Pick** to click a colour anywhere on screen, such as in the viewer. A neutral colour has no hue, so it cannot be picked. **Protect Neutrals** (default 0.1) reduces the effect on nearly grey colours, where hue is unstable. **Reset all curves** returns every curve to flat. Hue Correct is a good way to suppress green or blue spill after a key: pull **Hue → Saturation** or **Hue → Green** down around the backing's hue. ## Brilliance and Structure **Brilliance** (−1 to 1) works on local lighting. Positive values reveal detail in shadows and highlights; negative values deepen shadows and simplify lighting for a moodier look, without glow. Strong edges are preserved. **Structure** (the node for local contrast) changes contrast within regions instead of across the whole frame. **Amount** (−1 to 1) simplifies local lighting when negative, which helps tame a crunchy HDR look, and adds punch when positive. **Scale** sets the size of structures it targets, and **Fine Detail** (default 0.85) protects fine texture while larger structures change. Both nodes also have **Mix**, a **Mask** input and **Channels**, like the other effects. ## Limit a correction to part of the image Most correction nodes accept a **Mask** input with **Mask Source**, **Invert Mask** and **Mask Mix**, and a **Channels** control that chooses which RGBA channels are written. See [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/). ## Frequently asked questions ### Do colour nodes work in linear or display values? Inside a colour-managed compositor they work on scene-linear working-space values. In the ordinary Graph, most image effects decode the sRGB image to linear for the operation and encode the result back to sRGB. See [Color management](https://docs.nodespell.com/edit/color-management/). ### Why is Contrast pivot set to 0.18? 0.18 is the conventional scene-linear value for mid-grey, so contrast expands around a mid-grey that matches a linear working space. ### Can I animate a grade? Yes, inside a video Compositor. Numbers, colours, switches and menus can all be keyed; see [Keyframes](https://docs.nodespell.com/edit/keyframes/). ### Why does the Curves histogram stop at 1? The histogram is measured from an 8-bit preview of the input, so values above 1 are clipped in the display even though Curves itself processes them. ### Is there a Local Contrast node? The local-contrast node is named **Structure** in Nodespell. ## Related - [Color management](https://docs.nodespell.com/edit/color-management/): working spaces and view transforms. - [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/): limit a grade to an area. - [Keyframes](https://docs.nodespell.com/edit/keyframes/): animate grades over time. - [Viewers](https://docs.nodespell.com/edit/viewers/): check channels and read pixel values. - [Filters and effects](https://docs.nodespell.com/edit/filters-and-effects/): the full node map. --- # Color management, linear and OCIO in compositors > How Nodespell compositors manage color with OCIO: scene-linear working spaces, view transforms, input interpretation, EXR masters and what is not managed. Source: https://docs.nodespell.com/edit/color-management/ · Last updated: 2026-11-10 Color management in Nodespell is an OCIO pipeline built into compositors. New compositors convert every input into a scene-linear working space, process in floating point, show the result through a display and view transform, and deliver either a display-referred review image or a linear EXR master. If you know a Nuke show-LUT workflow, the model is the same, and it all runs in your browser. ## How does color flow through a compositor? A colour-managed compositor follows four stages, each with its own settings: | Stage | What happens | Where you set it | | ------------- | ----------------------------------------------------------------------------- | -------------------------------------------- | | Input | Each source is interpreted from its colour space into the working space | The compositor's **Inputs** | | Working space | Effects and merges run on scene-linear, floating-point values | Project setup, or a compositor override | | Viewing | The viewer applies a display, view and optional look for inspection only | Viewer toolbar and project review defaults | | Result | The compositor writes a review image or video "as viewed", or a linear master | **Result** in the compositor's colour status | Viewing never changes pixels. Exposure, gamma and saturation you set in the viewer are personal, stored in your browser, and never baked into results. ## What are the defaults? Every new compositor (Image Compositor and Compositor) is colour-managed and follows the project's colour setup. The project setup starts on **Standard**. | Setup | Configuration | Working space | Viewer display and view | | ---------------------- | ---------------------------------- | ------------------------------------- | ------------------------------------ | | **Standard** (default) | Built-in ACES 1.3 CG configuration | Linear Rec.709 (sRGB) | sRGB - Display, Un-tone-mapped | | **ACES** | Built-in ACES 1.3 CG configuration | ACEScg | sRGB - Display, ACES 1.0 - SDR Video | | **Studio config** | Your imported OCIO configuration | A scene-linear space from that config | From that config | Working precision defaults to half float; full float is available. Standard keeps the familiar sRGB look of ordinary images while you work in linear light. ACES gives a filmic tone-mapped view and an ACEScg working space for wide-gamut work. ## Set up colour for a project 1. Open the **Edit** menu and choose **Project Colour Management**. 2. Choose a setup: **Standard**, **ACES** or **Studio config**. Changing the setup asks you to confirm, because every compositor that follows the project updates immediately. 3. Optionally turn on a **Show look**: a client LUT or CDL applied to viewing and review results, never to master EXRs. Set **LUT output** to **Creative** (the view transform runs after it) or **Display-ready** (the LUT already includes the display transform). 4. Check the master settings: **Master EXR space** (the working space by default) and precision. 5. Under **Advanced**, set context variables, such as `SHOT`, if your configuration uses them. Only the project owner can save project colour settings. **Use these settings for new project compositors** applies the current settings to compositors created later without changing existing ones. To import a studio configuration, use **Import OCIO configuration** and choose a ZIP containing one `.ocio` file and its LUT files, keeping their relative folders. The maximum size is 64 MB. Paths must stay inside the bundle; configurations that need a machine path or an external file must be made portable first. ## Override colour for one compositor Right-click a compositor and choose **Colour settings…**, or use the colour status in its viewer. A compositor follows the project until you change its configuration or working space, which pins a custom pipeline for that compositor only. **Use project settings** reconnects it to the project. Input assignments and OCIO graph nodes always stay local to the compositor. ## Interpret inputs correctly When a source enters the compositor, Nodespell converts it into the working space. | Source | Automatic behaviour | | --------------- | ------------------------------------------------------------------------------ | | PNG, JPEG, WebP | Uses the embedded colour profile; standard or untagged images are read as sRGB | | EXR | Uses supported colour metadata; otherwise you must assign a space | | Video | Uses supported video colour tags; ambiguous sources need an assignment | In the compositor's **Inputs**, you can override any source's space, choose **Raw / data** to bypass colour transforms for depth, normals, motion vectors and other data, and mark that **Source RGB is premultiplied by alpha**. For correctly tagged SDR camera or video media, **Video Rec.709 signal (scene-linear decode)** reverses the Rec.709 video transfer to scene-linear. A source that needs a decision shows how many inputs need a colour space. Some video cannot enter a colour-managed compositor, such as rotated video, non-square pixels or HLG HDR without a manual choice. The compositor explains the problem and offers fixes, including **Turn off colour management for this compositor**, which you can undo. ## Convert inside the graph with OCIO nodes Colour-managed compositors add OCIO nodes to the Tab menu. Their choices are saved on the node, so copy, paste, undo and snapshots carry them. | Node | What it does | | --------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **OCIO ColorSpace** | Converts pixels between spaces in the compositor's configuration. **From** set to Automatic uses the frame's declared space; an explicit From reinterprets the pixels. **To** can return to the working space | | **OCIO Assign Colourspace** | Changes the declared space without changing pixel values. Use only when an interpretation is wrong | | **OCIO File Transform** | Applies an imported LUT, CDL, CLF or CTF | | **OCIO Look** | Applies a named look from the configuration | | **OCIO Named Transform** | Applies a named transform from the configuration | Import files for **OCIO File Transform** in the compositor's colour window, under **Transforms**. Files must be CUBE, 3DL, CLF, CTF, CC, CCC or CDL, up to 32 MB each. > **Caution:** An OCIO ColorSpace node can output a space other than the working space for inspection. Convert back to the working space before any effect, Merge or Compositor Output. A Merge whose inputs are in different spaces fails instead of guessing. ## View transforms and the viewer The compositor viewer shows the result through the selected **Display** and **View transform**, plus any look. The viewer toolbar adds personal **Exposure**, **Gamma** and **Saturation**, a show-look toggle, and **Bypass** to inspect raw values where the display has a raw view. The browser draws on an sRGB surface, so the viewer shows display signal values clipped to 0–1. Choosing another display target lets you inspect its signal values; it is not calibrated wide-gamut or HDR monitoring. ## Choose what the compositor delivers | Compositor | **Result** option | Output | | ---------------- | ---------------------------------------- | ------------------------------------------------------------------------------------------ | | Image Compositor | **Review · as viewed (PNG)** (default) | Display-referred PNG with the review display and view, without personal viewer adjustments | | Image Compositor | **Master · linear EXR** | Scene-linear EXR in the master space, half or full float, never with the show look | | Compositor | **Review · as viewed (video)** (default) | Display-referred video | | Compositor | **HDR review · 10-bit WebM PQ** | Rec.2100 PQ video, with the built-in configuration on supported devices only | A video Compositor can also write a linear EXR image sequence to a folder with **Export EXR sequence…** in desktop Chrome or Edge. Choose an empty folder. When a composite feeds an AI generator, use a review result: generators expect ordinary display images, not linear EXR data. ## What is and is not colour-managed | Area | Managed? | | ------------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Inside a colour-managed compositor | Yes: inputs, working space, viewer and results | | Older compositors saved before colour management | No, until you choose **Enable linear OCIO workflow**. They show **Working Precision** (8-bit, 16-bit float, 32-bit float), **Output Precision**, **View Transform** (Auto, LTX ACES SDR, sRGB, Rec.709, Raw linear) and **View Exposure** instead | | Image effect nodes in the ordinary Graph | No OCIO. Most decode the sRGB image to linear for the operation and encode back to sRGB | | **Color Space** node | Fixed conversions between sRGB, Linear sRGB / Rec.709, Rec.709, Display P3, Linear Display P3, Rec.2020, Linear Rec.2020, ACEScg, HSV, HSL, LAB / CIE L\*a\*b\* and CIE XYZ D65, with optional **Clamp Output**. Inside OCIO compositors, prefer OCIO ColorSpace | | Design panel image viewers outside compositors | Show stored image values; exposure, gamma and saturation are display filters only | | Fullscreen viewer for EXR files | Its own **Tone mapping** choices for display | When you enable OCIO on an older compositor that uses **Color Space** or **Apply LUT** nodes, Nodespell asks you to migrate them first and can migrate exact matches automatically. ## Frequently asked questions ### Does Nodespell use OCIO? Yes. Colour-managed compositors run OCIO 2 in the browser, with a built-in ACES 1.3 CG configuration or your own imported configuration. ### What working space do new compositors use? Linear Rec.709 (sRGB) in half float, with an sRGB display and un-tone-mapped view, unless the project uses the ACES or a studio setup. ### Is the viewer exposure baked into my output? No. Viewer exposure, gamma and saturation are personal display settings. Review results use the project's display and view without them, and masters stay linear. ### Can I monitor HDR in Nodespell? No. The viewer shows SDR display values on an sRGB surface. You can deliver an HDR review video or linear EXR masters for review elsewhere. ### Why does my video need a colour choice? Its colour tags are missing, ambiguous or unsupported, such as HLG. Assign a space in the compositor's Inputs, re-export the video, or turn off colour management for that compositor. ## Related - [Compositing](https://docs.nodespell.com/edit/compositing/): compositor types and Merge. - [Color correction](https://docs.nodespell.com/edit/color-correction/): grading nodes. - [Depth and defocus](https://docs.nodespell.com/edit/depth-and-defocus/): depth data as Raw input. - [Viewers](https://docs.nodespell.com/edit/viewers/): channels, exposure and pixel readout. - [Export](https://docs.nodespell.com/edit/export/): getting results out. --- # Depth maps, defocus and depth atmosphere > Create depth maps and normals in Nodespell, then refocus with Defocus and ZDefocus, shape bokeh, and add haze and fog with Depth Atmosphere. Source: https://docs.nodespell.com/edit/depth-and-defocus/ · Last updated: 2026-11-10 Depth effects in Nodespell turn a depth map, a grayscale image where brightness means distance, into lens blur and atmosphere. You create the depth map with a depth model node, then use **ZDefocus** to refocus the shot, **Defocus** for uniform lens blur, and **Depth Atmosphere** for haze, fog and diffusion that grow with distance. ## Create depth maps and normals Depth and normal model nodes run as tasks on Nodespell and return images you can feed into effects. | Node | Input | Outputs | Notes | | --------------------------- | ----- | -------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------ | | **Depth Anything V3 Image** | Image | **Result** | Grayscale depth: near is white, far is black. **Output format** PNG (default), JPEG or WebP; use PNG to avoid compression | | **Depth Anything V3 Video** | Video | **Result** | Temporally consistent depth video at the source resolution, keeping the audio, up to 30 fps; clips up to 10 minutes | | **Facet Depth V1** | Image | **Depth map (PNG)**, **Depth (EXR)**, **Raw depth (NPY)** | EXR and NPY keep floating-point depth. **Scale** (0.2–2) resizes the input, up to 12 megapixels. **Ignore transparent objects** estimates depth behind glass | | **Facet Normals V1** | Image | **Normal map (PNG)**, **Normals (EXR)**, **Raw normals (NPY)** | Camera-space normals, not tangent-space texture maps | All of these give relative depth, not distance in metres. When you bring depth or normals into a colour-managed compositor, assign them **Raw / data** in its Inputs so no colour transform changes the values. See [Color management](https://docs.nodespell.com/edit/color-management/). ## Defocus or ZDefocus? | | **Defocus** | **ZDefocus** | | ----------- | ----------------------------------------------- | --------------------------------------------- | | Blur amount | The same everywhere (**Radius**, default 16 px) | Varies with depth from the **Depth** input | | Use for | Out-of-focus elements, backgrounds, bokeh looks | Refocusing a shot, rack-focus, depth of field | | Extra input | **Kernel** (custom aperture) | **Depth** and **Kernel** | Both share the same apertures, quality setting and highlight controls, and both work in the Graph and in compositors. ## Refocus a shot with ZDefocus 1. Connect the image to **ZDefocus** and the depth map to its **Depth** input. 2. In the **Depth** tab, set **Depth Channel** (default **Luminance**) and **Depth Encoding**: **White Near** (default, as Depth Anything and Facet output) or **Black Near**. Use **Depth Min** and **Depth Max** to map your values to the 0–1 range. 3. Set **Output** to **Depth View** to check the normalised depth, then **False Colour Focus** to see what is in focus. 4. Set **Focus Depth** (0–1, default 0.5) to the depth that should stay sharp, and **Focus Range** (default 0.04) for how much depth around it stays sharp. 5. In the **Defocus** tab, set **Max Radius** (default 24 px) and the **Near Blur** and **Far Blur** strengths (default 1). 6. Set **Output** back to **Result**. | **Output** | Shows | | ---------------------- | --------------------------------------------- | | **Result** | The refocused image | | **Depth View** | Normalised depth | | **Blur View** | Blur amount per pixel | | **False Colour Focus** | Near, focused and far areas in colour | | **Kernel** | The aperture shape at its size on the picture | | **Layer Setup** | How depth is split into blur layers | ZDefocus renders depth in layers. **Automatic Layers** (on) chooses spacing from the radius and quality; turn it off to set **Layers Per Side** (default 20) and **Layer Curve** (0 is even; positive packs layers near focus). Foreground blur spreads beyond object edges. **Edge Anti-aliasing** smooths the silhouettes of focused objects against the depth map, with **Strength**. ## Choose a bokeh shape **Aperture** sets the lens shape for both nodes. New nodes use **Disc**; nodes saved earlier keep their blade aperture. | **Aperture** | Extra controls | | ------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Disc** | **Edge softness**, **Rim strength**, **Rim width** | | **Blades** | **Bokeh Sides** (3–16, default 7), plus the Disc controls | | **Star** | **Bokeh Sides** | | **Heart** | None | | **Ring** | **Hole size** (default 0.5) | | **Image** | Uses the picture on the **Kernel** input: **Kernel from** (**Auto**, **Brightness**, **Alpha**), **Kernel centre** (**Shape bounds** or **Image frame**), **Noise floor** | **Aspect** stretches the aperture, for an anamorphic look, and **Rotation** turns it. For an image kernel, connect a white shape on black, or a shape cut out of transparency; raise **Noise floor** to clean a photographed kernel. Set **Output** to **Kernel** to see the shape the blur uses. **Highlight Threshold** and **Highlight Gain** (default 0, off) boost bright areas before blurring so bokeh balls bloom. This is a creative adjustment for SDR footage; HDR values above 1 already produce bright bokeh. **Quality** is **Fast** (default, smaller working resolution) or **High quality**; preview and export use the same setting. During playback in a compositor, heavy defocus may show a half-resolution preview that is replaced by full quality when you pause. ## Add haze and fog with Depth Atmosphere **Depth Atmosphere** builds haze, dense fog and diffusion from a depth map. Connect **Image** and an aligned **Depth**; **Mask** is optional. It is a screen-space approximation: it cannot see hidden geometry, cast volumetric shadows or handle several depth layers per pixel. | Tab | Controls | | -------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Atmosphere** | **Start** (0.05) and **End** (1) distances, **Air Density** (0.35, up to 20), **Maximum Opacity** (0.9), **Air Colour**, **Falloff**, **Falloff Power** and **Falloff Curve** | | **Grade** | **Exposure** (stops), **Contrast** and **Saturation**, applied more with distance | | **Diffusion** | **Image Diffusion** softens all distant detail; **Highlight Bloom** (0.35) spreads light from bright areas, with **Highlight Threshold**, **Soft Knee**, **Radius** (80 px), **Blur Layers**, **Layer Growth** and **Tail Falloff** | | **Depth** | **Depth Channel**, **Depth Encoding** (**White Near** default), **Decode sRGB Depth**, **Depth Min**, **Depth Max**, **Mask Softness** (1 px) and **Edge Anti-aliasing** (1) | **Falloff** is **Exponential** (default), **Linear** or **Custom**. With **Custom**, draw the distance ramp as a curve: its endpoints are fixed, points can be smooth or linear, and you can open a larger editor. **Falloff Power** shapes the ramp before the falloff is applied. Turn on **Decode sRGB Depth** only for a depth map stored as an sRGB image; leave it off for numeric depth. Set **Output** to inspect parts of the effect: **Result**, **Atmosphere Only**, **Distance Influence**, **Transmission**, **Highlight Bloom**, **Diffused Image**, **Fog Mask** or **Diffusion Mask**. ## Frequently asked questions ### Which depth model should I use? Use Depth Anything V3 Image for a quick grayscale map and the Video node for moving shots. Use Facet Depth V1 when you need floating-point depth in EXR or NPY. ### Why is my focus inverted? Your depth map may use black for near. Set **Depth Encoding** to **Black Near**. ### Are depth values in real-world units? No. These models produce relative depth. Use **Focus Depth**, **Depth Min** and **Depth Max** to work in the normalised 0–1 range. ### Can I animate the focus? Yes, inside a video Compositor. Key **Focus Depth** for a rack focus; see [Keyframes](https://docs.nodespell.com/edit/keyframes/). ### Do Defocus and Depth Atmosphere use credits? They run on your GPU in the browser. The depth and normal model nodes run as tasks on Nodespell. ## Related - [Recipe: depth refocus and atmosphere](https://docs.nodespell.com/recipes/depth-refocus-and-atmosphere/): a worked example. - [Filters and effects](https://docs.nodespell.com/edit/filters-and-effects/): other blurs and filters. - [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/): depth maps as mattes. - [Color management](https://docs.nodespell.com/edit/color-management/): keep depth data raw. - [Keyframes](https://docs.nodespell.com/edit/keyframes/): animate focus. --- # Image filters and effects node map > A guided map of Nodespell image effect nodes: Denoise, blurs, glow, sharpening, generators, distortion, text, time and motion nodes, and where each works. Source: https://docs.nodespell.com/edit/filters-and-effects/ · Last updated: 2026-11-10 Image effect nodes in Nodespell are built-in nodes that filter, generate, distort or retime pictures on your GPU, in the browser, without using an AI model. Most work both in the ordinary Graph on still images and inside [compositors](https://docs.nodespell.com/edit/compositing/); time and motion nodes need a video Compositor. This page maps every family so you can find the right node. ## Where does each node work? | Where | What you can use | | --------------------------- | ----------------------------------------------------------------------------------- | | Graph (outside compositors) | Most filters, colour nodes and generators, on still images | | Image Compositor | The same, plus compositor-only nodes such as Text, Roto, STMap and Channel Displace | | Compositor (video) | Everything above, plus time and motion nodes | Inside a compositor the Tab menu lists only nodes that work there. To try an effect before adding it, select an image node and browse effects in the [Tab menu](https://docs.nodespell.com/graph/add-nodes/), which shows a preview against the source. ## Controls most effects share | Control | What it does | | ------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------- | | **Mix** | Blends the result with the original image | | **Mask** input with **Mask Source**, **Invert Mask**, **Mask Mix** | Limits the effect to part of the image; see [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/) | | **Channels** | Chooses which RGBA channels the node writes; the rest pass through | | **Edges** | How blurs and distortions sample past the frame edge: **Clamp**, **Black**, **Wrap** or **Mirror** (options vary by node) | ## Clean up with Denoise **Denoise** removes noise and grain while keeping detail. It measures the noise, removes it with non-local means and a coarse pass for blotches, and puts back luma detail that stands above the noise. | Group | Controls | | ------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | **Strength** | **Auto strength** (on) sets strength from the measured noise. Turn it off to set **Luminance** (0.7), **Colour** (0.75) and **Smooth areas** (0.6). **Blotchy noise** (0.4) targets low-frequency mottling. **Quality** is **Fast (7 × 7 search)**, **Balanced (11 × 11 search)** (default) or **Best (17 × 17 search)** | | **Detail** | **Detail preservation** (0.8), **Detail recovery** (0.2), **Sharpen** (0) and **Grain re-injection** (0), which adds fresh fine grain to avoid a plastic look | | **Noise** | **Source**: **Auto (whole image)** uses the smoothest areas, **Measure region** measures a flat area you place, **Manual** uses **Luma noise** and **Colour noise** levels (standard deviation in 8-bit levels) | With **Measure region**, a **Noise region** box appears on the viewer. Place it on flat ground with no edges, texture or shadows; 80 × 80 pixels or more measures best. ## Filters | Node | What it does | Typical use | | ----------------------------- | ------------------------------------------------------------------------------------------ | ------------------------------------------------- | | **Blur** | Separable blur with **Quadratic**, **Triangle** or **Box** filter, **Radius** up to 200 px | General softening | | **Directional Blur** | Blur along an **Angle** over a **Distance** | Speed streaks, wind | | **Exponential Blur** | Blends several blurs of growing radius | Broad VFX-style softening and glow | | **Glow** | Thresholded glow with tint and **Blend** (**Screen**, **Add**, **Over**) | Light sources, emissive elements | | **Defocus**, **ZDefocus** | Lens blur with bokeh shapes, flat or depth-driven | See [Depth and defocus](https://docs.nodespell.com/edit/depth-and-defocus/) | | **Depth Atmosphere** | Haze, fog and diffusion from depth | See [Depth and defocus](https://docs.nodespell.com/edit/depth-and-defocus/) | | **Denoise** | Detail-preserving denoise and degrain | Cleaning AI or camera noise | | **Brilliance**, **Structure** | Local lighting and local contrast | See [Color correction](https://docs.nodespell.com/edit/color-correction/) | | **Sharpen** | Sharpening in LAB to avoid colour artefacts | Quick crispness | | **Unsharp Mask** | Sharpening with **Amount**, **Radius** and **Threshold** | Controlled sharpening | | **Grain** | Signal-dependent film grain with size and per-channel strength | Matching plates, adding texture | | **Opacity** | Changes alpha | Fades, transparency | | **Edge Detect** | Black-and-white edge map | Line mattes, stylised looks | | **Pixelate**, **Posterize** | Mosaic blocks; fewer colour levels | Stylised looks | | **Chromatic Aberration** | Colour fringing from a centre point | Lens realism | | **Vignette** | Darkens or colours the edges | Framing | ## Generators and text | Node | What it does | | -------------------------- | ------------------------------------------------------------------------------------------------------- | | **Gradient** | Linear, Radial, Angular or Reflected gradients with extra colour stops | | **Noise** | Value, Perlin-like, Fractal or Cellular noise, animatable with **Evolution** | | **Checkerboard** | Checker pattern for alignment and distortion checks | | **Rectangle**, **Ellipse** | Antialiased shapes with stroke and edge fade, generated or overlaid; drag them in the compositor viewer | | **Fill** | Solid colour or gradient, or fill behind transparent areas | | **Create Image** | A new solid, checkerboard or UV-map image at a preset or custom size (Graph only) | | **STMap Generator** | An identity ST map (red is X, green is Y) | | **Text** | Styled text with font, wrapping, stroke, shadow and position (compositors only) | ## Distortion and layout | Node | What it does | | -------------------- | ----------------------------------------------------------------------------------- | | **Transform** | Scale, move, rotate and skew; see [Transforms and crop](https://docs.nodespell.com/edit/transforms-and-crop/) | | **Lens Distortion** | Barrel or pincushion distortion | | **STMap** | Remaps pixels from UV coordinates in a map image (compositors only) | | **Channel Displace** | Offsets pixels using two channels of a displacement image (compositors only) | | **Tile**, **Mirror** | Repeat or flip the image | | **Contact Sheet** | Lays several images out in a grid (Graph only) | ## Time and motion (video Compositor only) | Node | What it does | | ---------------- | --------------------------------------------------------------------------------------------------------------------------------------------------- | | **Time Offset** | Samples the stream earlier or later, in frames and seconds | | **Freeze Frame** | Holds one source frame | | **Reverse Time** | Plays a frame range backwards | | **Retime** | Changes speed, fits a target duration, or follows an animated source-frame curve; **Sampling** is **Nearest**, **Linear Blend** or **Optical Flow** | | **Optical Flow** | Calculates a motion-vector map from neighbouring frames | | **Motion Blur** | Flow-driven motion blur with **Shutter Angle** (default 180) | | **Camera Shake** | Procedural shake with presets **Handheld**, **Impact**, **Rumble** and **Micro Jitter** | ## Custom shaders **Shader Effect** applies your own realtime GLSL fragment shader to an image or compositor frame, with up to four passes. You can also ask the [Agent](https://docs.nodespell.com/agent/) to create or edit one with **Edit Shader Effect with Agent...** from the node's actions. ## Frequently asked questions ### Do effect nodes cost credits? No AI model is involved. Effect nodes render on your GPU in the browser. ### Why can't I find Retime or Camera Shake? Time and motion nodes appear only inside a video Compositor. Convert an Image Compositor to a video Compositor to use them. ### Can effects process video in the ordinary Graph? Graph image effects process still images. Put video inside a Compositor to apply effects to every frame. ### What does Denoise measure in Auto mode? Auto measures noise in the smoothest areas of the whole image and sets strength from it. Choose **Measure region** when the picture has few flat areas. ## Related - [Compositing](https://docs.nodespell.com/edit/compositing/): build multi-node effect graphs. - [Color correction](https://docs.nodespell.com/edit/color-correction/): grading nodes. - [Depth and defocus](https://docs.nodespell.com/edit/depth-and-defocus/): lens and atmosphere effects. - [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/): limit effects. - [Keyframes](https://docs.nodespell.com/edit/keyframes/): animate effect parameters. --- # Animate parameters with keyframes > Animate numbers, colours, switches and menus in a Nodespell video Compositor with keyframes, the dopesheet, the curve editor and easing. Source: https://docs.nodespell.com/edit/keyframes/ · Last updated: 2026-11-10 Keyframes in Nodespell animate node parameters over time inside a video **Compositor**. You set a key at the playhead from the parameter's row in Parameters, Nodespell interpolates between keys, and the dopesheet and curve editor let you move keys and shape their easing. Numbers, colours, switches and menus can all be animated. ## What can be animated? Keyframes work on nodes inside a video Compositor, which has a timeline and playhead. An **Image Compositor** has no timeline, so its rows show **Cannot animate**; convert it to a video Compositor to animate. Nodes in the ordinary Graph, outside compositors, do not animate. | Parameter type | Animates as | Between keys | | -------------------------------- | -------------------------------------------------- | ------------------------- | | Numbers (sliders, number fields) | A curve | Interpolated, with easing | | Colours | Red, green and blue channel curves, keyed together | Interpolated per channel | | Switches (on/off) | Step keys | Holds until the next key | | Menus (two or more options) | Step keys | Holds until the next key | | Text, file and asset choices | Not animatable | No key control is shown | Roto shapes have their own shape keys; see [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/). ## Set and remove a key 1. Go inside a video Compositor and select a node, so its parameters appear in [Parameters](https://docs.nodespell.com/graph/parameters/). 2. Move the playhead to the frame you want. 3. Click the diamond beside the parameter, or put the keyboard focus in the row and press `K`. The diamond fills to show **Key on this frame**. 4. Move the playhead and change the value. Set another key at the new frame. 5. To remove a key, go to its frame and click the filled diamond or press `K` again. The arrows on either side of the diamond jump the playhead to the **Previous key** and **Next key**. Keying a colour keys all three channels in one step, so a single Undo removes it. ## Read the key control | Diamond | Meaning | | -------------------------------------- | -------------------------------------------------------------------------- | | Outline | **Not animated** | | Outline with a dot | **Animated, no key on this frame** | | Filled (square for switches and menus) | **Key on this frame** | | Half filled | **Some channels are keyed on this frame. Click to key the rest** (colours) | | Amber | **Changed since the last key, not keyed** | | Dashed | **Cannot animate** | Parameter rows follow the playhead: as you scrub or play, each animated row shows its value at the current frame. Changing a value on a key updates that key. Changing an animated value between keys shows it in amber and previews it, but it is not saved into the animation until you set a key. Nodespell has no auto-key. ## Edit keys in the dopesheet Open the dopesheet from the diamond button with the key count in the compositor viewer's header. The **Animation** panel lists every animated parameter with a lane of keys, grouped by node. Colours show one summary lane that expands into **Red**, **Green** and **Blue**; switches and menus show each held value as a bar labelled with its value, such as **On** or **Off**. | Action | How | | ------------------ | --------------------------------------------------------------------------------------- | | Add a key | Double-click a row at the frame you want | | Select keys | Click; `Shift`, `Cmd` or `Ctrl`+click to add to the selection; `Cmd/Ctrl+A` selects all | | Move keys | Drag them, or press the left and right arrow keys (1 frame; 10 with `Shift`) | | Edit a key exactly | Select it and set **Frame**, **Value** and **Easing out of the key** | | Delete keys | `Delete` or `Backspace`, or **Delete keys** | | Deselect | `Esc` | | Clear everything | **Clear all keys**, then click again to confirm | | Jump between keys | **Previous key** and **Next key** | ## Shape easing in the curve editor Switch the Animation panel from **Dopesheet** to **Curves** to edit animation curves. Switches and menus have no curve, because they change in steps; colours show one channel at a time. | **Easing** | Behaviour | | ------------------------------ | ----------------------------------------------------------------- | | **Smooth** | Eases through the key | | **Linear** | Constant speed to the next key | | **Hold** | Keeps the value until the next key | | **Handles** (**Custom curve**) | Bezier handles you drag; **Joined** or **Split**, and **Flatten** | **More easings** offers **Ease in**, **Ease out** and **Ease in-out** presets, alongside **Reset key**. In the curve editor, `Cmd/Ctrl+C` and `Cmd/Ctrl+V` copy and paste keys, `Delete` removes them, `F` fits the view, and the arrow keys nudge the selected key in frame and value. ## Frequently asked questions ### Why does my parameter say Cannot animate? The node is inside an Image Compositor, which has no timeline. Use **Convert to Video Compositor** from the compositor's actions to animate it. ### Can I keyframe nodes outside a compositor? No. Keyframes belong to a video Compositor's timeline. Put the nodes inside a Compositor to animate them. ### Why did my change disappear when I moved the playhead? The parameter was already animated and you changed it between keys without setting a key. The amber diamond warns about this; press `K` to keep the change. ### Can a menu, like a blend mode, change over time? Yes. Menus and switches take step keys that hold their value until the next key. ## Related - [Compositing](https://docs.nodespell.com/edit/compositing/): Image Compositor and Compositor. - [Parameters](https://docs.nodespell.com/graph/parameters/): editing values. - [Expressions](https://docs.nodespell.com/graph/expressions/): drive values with formulas instead of keys. - [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/): animate Roto shapes. - [Timeline](https://docs.nodespell.com/edit/timeline/): editorial sequencing of clips. --- # Transform, resize and crop images > Move, scale and rotate images in Nodespell with Transform, change size with Resize, and crop with on-image handles, aspect presets and reformat. Source: https://docs.nodespell.com/edit/transforms-and-crop/ · Last updated: 2026-11-10 Nodespell has three nodes for geometry: **Transform** moves, scales, rotates and skews an image inside its existing frame; **Resize** changes the image's pixel dimensions to a format, scale or reference; and **Crop** cuts a region out, either keeping the frame size or reformatting to the crop. All three work in the Graph and inside compositors. ## Which node changes the image size? | Node | Output size | Use for | | ------------- | -------------------------------------------------------- | ------------------------------------------------------- | | **Transform** | Same as the input | Positioning a layer, rotating, scaling inside the frame | | **Resize** | A format, custom size, scale or reference image | Matching a delivery size or another image | | **Crop** | Same as the input, or the crop size with **reformat** on | Removing edges, reframing | Every other image effect outputs the same size as its input, so a chain of filters keeps your source resolution. In the Design panel, **Preview resolution** only changes the preview; see [Viewers](https://docs.nodespell.com/edit/viewers/). ## Move, scale and rotate with Transform | Control | Range and default | Notes | | -------------------------------- | ------------------------------------------------------------------------- | --------------------------------------------- | | **Scale X**, **Scale Y** | 0.01–10, default 1 | | | **Translate X**, **Translate Y** | −2 to 2, default 0 | In frame widths and heights; Y moves up | | **Rotate** | −360 to 360 degrees | Positive turns clockwise on screen | | **Skew X**, **Skew Y** | −89 to 89 degrees | | | **Pivot X**, **Pivot Y** | 0–1, default 0.5 (centre) | The point scale and rotation happen around | | **Filter** | **Cubic** (default), **Mitchell**, **Lanczos 3**, **Linear**, **Nearest** | Resampling quality; Nearest keeps hard pixels | Inside a compositor, select a Transform node to get an on-image control in the viewer: a ring that stands in for the image. | Gesture | Result | | --------------------------------- | ----------------------------------------------------------------- | | Drag inside the ring | Move the image | | `Option/Alt`+drag inside the ring | Move the pivot without moving the image | | Drag a square on the ring | Scale about the pivot; hold `Shift` to scale along that axis only | | Drag the round grip | Rotate about the pivot; hold `Shift` to snap to 15° steps | The amber target marks the pivot. All gestures use the primary button, so they work on a trackpad. ## Change size with Resize | **Target** | Size comes from | | ----------------------- | ----------------------------------------------------------------------------------------------------------------------------------------- | | **Automatic** (default) | The **Reference** input when connected; otherwise the compositor's canvas, or the image's own size outside a compositor | | **Composition** | The compositor canvas | | **Preset** | **Format**: HD · 1920 × 1080, UHD · 3840 × 2160, Portrait · 1080 × 1920, Square · 1080 × 1080, DCI 2K · 2048 × 1080, DCI 4K · 4096 × 2160 | | **Custom** | **Width** and **Height**, 1–8192 pixels | | **Reference** | The dimensions of the image on **Reference** (only its size is used) | | **Scale** | **Scale**, 0.01–16 times the input | When the shape changes, **Fit** decides what happens: **Contain** (default) pads with transparency, **Cover** crops the overflow, **Stretch** changes proportions, and **No Resize** only crops or pads. **Anchor** picks which part of the image stays aligned when padding or cropping. **Filter** is **Smooth** or **Nearest**. ## Crop with on-image handles **Crop** stores the region in normalised coordinates from 0 to 1, with Y measured up from the bottom, as in Nuke. **Mode** is **Edges (L/R/T/B)** (default; left 0, right 1, top 1, bottom 0) or **Center (W/H/X/Y)**. 1. Add **Crop** after an image and select it. The Design panel shows the crop editor over the input image. 2. Drag the edges or corners. Hold `Shift` to keep the shape, or `Alt` to crop from the centre. Drag inside the crop to move it. 3. Pick a shape: **Free**, **Original**, **1:1**, **4:3**, **3:2**, **16:9**, **21:9** or **Custom**. **Swap orientation** turns the ratio portrait. For Custom, type a ratio and select **Set**. 4. Type an exact **Width** and **Height** in pixels if you need them; the field shows the source size beside them. 5. Use the arrow keys to nudge the crop, with `Shift` for larger steps. **Reset** returns to the full image. The same handles, shape presets and size fields appear in a compositor's viewer when you select a Crop node there, and the preview follows as you drag. You can also open **Edit Crop** from the node, or crop from the fullscreen viewer: choose **Crop** in its tools, adjust, and apply to add a Crop node after the image you are viewing. Nothing is created until you apply. ## Keep or change the frame with reformat The **Reformat** switch (off by default) decides the output size: - **Off**: the output keeps the input size, and pixels outside the crop become transparent. Use this inside composites when the layer must stay registered to the plate. - **On**: the output is resized to the crop bounds. Use this to deliver the cropped region as its own image, for example before sending it to a generator. ## Frequently asked questions ### Why is my cropped image still the original size? **Reformat** is off, so the crop keeps the frame and makes the outside transparent. Turn **Reformat** on to output only the cropped region. ### How do I match one image's size to another? Connect the other image to Resize's **Reference** input. With **Target** on **Automatic** or **Reference**, Resize uses its dimensions. ### Does Transform change the image resolution? No. Transform works inside the input's frame, so parts moved outside are cut off. Use Resize to change the pixel dimensions. ### Do downstream generators get my transformed image? Yes. Generators downstream use the current Transform, Resize and Crop results when they run. ## Related - [Compositing](https://docs.nodespell.com/edit/compositing/): canvas size and layering. - [Viewers](https://docs.nodespell.com/edit/viewers/): zoom and inspect at full detail. - [Keyframes](https://docs.nodespell.com/edit/keyframes/): animate a Transform. - [Filters and effects](https://docs.nodespell.com/edit/filters-and-effects/): distortion nodes. - [Export](https://docs.nodespell.com/edit/export/): delivery sizes and formats. --- # Inspect results in the Design panel viewers > Inspect Nodespell results in the Design panel: zoom, colour channels, pixel values, two-image comparison, preview resolution, text and fullscreen viewers. Source: https://docs.nodespell.com/edit/viewers/ · Last updated: 2026-11-10 The Design panel viewers in Nodespell show the selected node's image, video or text so you can inspect it closely. Image and video viewers share the compositor viewer's tools: zoom and pan, channel isolation, display exposure, a colour readout and two-image comparison. Text has its own viewer with search, wrapping and Markdown. ## What does the Design panel show? The Design panel follows your selection in the Graph. Open it from the panel menus, or right-click a node and choose **Open Design**. | Selection | Viewer | | ------------------------------------ | ----------------------------------------------------------------------------- | | An Image asset | Image viewer | | An image effect node | Image viewer with an **Input** / **Output** toggle | | Any node with image or video outputs | Image or video viewer, with a tab for each output | | A Video asset | Video viewer: the player plus channels, display adjustments, zoom and readout | | A Text asset or text output | Text viewer | | Two nodes that output images | Image comparison | | A compositor or a node inside one | The compositor viewer, with timeline and colour controls | Re-running a node or switching output tabs keeps your zoom, channel and adjustments. ## Zoom and pan | Action | How | | -------------- | ------------------------------------------------------------------------------- | | Pan | Scroll with two fingers on a trackpad, or drag with the middle mouse button | | Zoom | Pinch on a trackpad, or hold `Cmd/Ctrl` and scroll; zoom centres on the pointer | | Zoom in or out | `+` and `-` | | Fit | `F` (in the video viewer, `0`) | | Actual size | `Shift+F` (in the video viewer, `1`) | | Zoom presets | The zoom menu: **Fit**, **Actual size**, 25%, 50%, 75%, 150%, 200%, 400% | Fit never enlarges a picture past 100%. Shortcuts work when the viewer has focus; click the picture first. ### Keep framing between shots The Design panel remembers your zoom and pan when you select another picture with the same aspect ratio, even at a different pixel size, so you can step through matching shots and look at the same detail. A picture with a different shape opens at Fit. ## Inspect channels, exposure and pixel values - **Channels:** the channel menu shows **RGBA**, **RGB**, **Red**, **Green**, **Blue**, **Alpha**, **Luminance** or **Saturation**. Press `R`, `G`, `B` or `A` to view a single channel. Video has no alpha channel. The channel choice is shared by all viewers. - **Display adjustments:** **Exposure**, **Gamma** and **Saturation** change only what you see. Drag a value, click to type, or double-click to reset. - **Readout:** the strip under the picture shows the colour and position under the pointer. X and Y are in the picture's own pixels, with Y counted up from the bottom edge, as in a compositor. Readouts sample the unadjusted pixel. - **Copy a colour:** right-click the picture and choose **Copy colour under cursor** as Hex, RGB, HSL or HSV. - **Background:** right-click and choose **Viewer background**: **Black**, **White**, **Light checkerboard**, **Dark checkerboard** or **Custom…**. It shows behind transparent pixels and never affects exports. On image effect nodes, press `I` or use **Input** / **Output** to flip between the node's input and its result. ## Change the preview resolution For image effect nodes, **Preview resolution** in the viewer bar sets how detailed the live preview is: **Use preference** (your saved preview size), **1:1 · Full resolution**, or 1:2, 1:3, 1:4, 1:6 and 1:8. The choice applies to image-processing viewers until you refresh the app, and it does not change saved projects or rendered results. Use **1:1 · Full resolution** to check fine detail, such as noise or sharpening. ## Compare two images 1. Select two nodes that output images, such as an original and a graded version. The first one you select is A, the second is B. 2. The Design panel switches to a comparison. Choose a mode from the mode menu. 3. Press `X` or use **Swap A and B (X)** to swap the sides. | Mode | Shows | | -------------------------------- | ----------------------------------------- | | Horizontal swipe, Vertical swipe | A divider you drag across the picture | | Side by side, Stacked | Both pictures next to or above each other | | Flip | A and B in turn | | Dissolve | A blend of the two | | Difference | A heatmap of where they differ | Both pictures are fitted into one common frame without cropping, so pan, zoom and the readout line up. The readout shows both colours at the pointer and their largest channel difference. Channels and display adjustments apply to the comparison too. To keep a comparison in the graph, use the **Compare Image**, **Compare Video** or **Compare Text** nodes, which show two connected inputs on the node with the same layouts. ## Read text outputs | Control | What it does | | ------------------------- | ---------------------------------------------------------------- | | **Raw** / **Markdown** | Shows text as written or rendered; `.md` files open as Markdown | | **Wrap** (`W`) | Wraps long lines | | **Lines** (`L`) | Shows line numbers | | Text size (`-`, `+`, `0`) | 75% to 200% | | **Find** (`Cmd/Ctrl+F`) | Highlights matches; `Enter` and `Shift+Enter` step, `Esc` clears | | **Copy all text** | Copies everything, even when a long file is shortened | Wrap, line numbers and size are remembered between files. Very long text shows its first 50,000 lines, and Find stops counting at 5,000 matches. ## Open the fullscreen viewer Select a node and press `Space`, or right-click it and choose **View Full Screen**. The fullscreen viewer shows the node's takes with **Compare (C)**, **Grid View (G)**, **Toggle info (I)** and **Close (Esc)**. EXR images get **Tone mapping** and **Exposure** controls for display. Its tools include **Crop**, which adds a Crop node when you apply. See [Takes](https://docs.nodespell.com/graph/takes/) for choosing results. 3D model results open in a fullscreen 3D viewer with **Reset Camera (H)**, material presets, **Download Model** and **Save Render**. ## Frequently asked questions ### Does viewer exposure change my image? No. Exposure, gamma, saturation and channel views are display-only. Readouts and results use the real pixel values. ### Why does my image look soft in the viewer? The preview may be running below full resolution. Set **Preview resolution** to **1:1 · Full resolution**, or zoom to **Actual size**. ### How do I pan without a middle mouse button? Scroll with two fingers on a trackpad. Pinch or `Cmd/Ctrl`+scroll zooms. ### Why is Y measured from the bottom? Nodespell follows compositing convention, where the origin is the bottom-left corner, so readouts match compositor coordinates. ## Related - [Takes](https://docs.nodespell.com/graph/takes/): browse and choose results. - [Color management](https://docs.nodespell.com/edit/color-management/): the compositor viewer's display and view. - [Compositing](https://docs.nodespell.com/edit/compositing/): what the compositor viewer shows. - [Transforms and crop](https://docs.nodespell.com/edit/transforms-and-crop/): on-image crop handles. - [Panels](https://docs.nodespell.com/reference/panels/): every panel in the workspace. --- # Edit video and audio in the Timeline > Use the Nodespell Timeline node to arrange video, image and audio clips on tracks, trim and mix them, and render a video back into your graph. Source: https://docs.nodespell.com/edit/timeline/ · Last updated: 2026-11-10 The Timeline in Nodespell is a node with a built-in multi-track editor. You place video, image and audio clips on tracks, trim and arrange them frame-accurately, mix the sound, and render one video. The render feeds the Timeline node's **Video** output, so the cut can continue through the rest of your graph. ## Add a Timeline and open the editor 1. In Graph, press `Tab` to open the [Tab menu](https://docs.nodespell.com/graph/add-nodes/), search for **Timeline**, and add it. 2. Connect the outputs you want to edit, such as generated videos, images or audio, to the Timeline node's **Assets** input. The input accepts many connections. 3. Right-click the node and choose **Edit Timeline**, or select it and open the [Design panel](https://docs.nodespell.com/edit/viewers/). The editor opens there. A new Timeline starts with one video track (**Video 1**), one audio track (**Audio 1**), a 1920×1080 sequence at 30 fps, snapping on, and linked audio on. The editor needs a current desktop browser. If required browser features are missing, it shows **Timeline editor needs a modern desktop browser** and asks you to switch to an up-to-date Chromium- or Firefox-based desktop browser. ## What is in the Timeline editor? | Area | What it does | | ------------------ | -------------------------------------------------------------------------------------------------- | | Top bar | Sequence summary (size, frame rate, length), preview status, **Render**, and the side panel toggle | | Viewer | Plays the sequence, with transport controls, preview volume and **Focus** for a larger viewer | | **Media** tab | Everything connected to the **Assets** input, ready to drag onto a track | | **Properties** tab | Settings for the selected clip or track, or the sequence when nothing is selected | | Toolbar | Tools, snapping, linked audio, split, delete, zoom, **Mixer** and **View** | | Tracks | Video and audio tracks with headers and clips | | Mixer | One strip per audio track plus **Master** | ## Add clips to tracks Open the **Media** tab. Each item shows where it came from and whether it is already used. - Drag an item onto a track, or use **Add to timeline**. - **Add all to timeline** places every item; **Add unused** places only items not yet on the timeline. - Filter the list with **All**, **Video**, **Audio** or **Not used**. - You can also drag Assets from the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/) into the Media area; they are added to the node's **Assets** input. Tracks are type-locked. Video and image clips go on video tracks; audio clips go on audio tracks, and a video's sound can go on an audio track. An image clip starts five seconds long. Another Timeline's output appears as **Nested**, and a Compositor's output appears as **Comp**. When a connected node produces a new take, the Media item shows **Changed upstream**. Choose **Update to new version** to swap the clips to the new take; trims and effects stay as they are. ## Arrange, trim and split clips | Tool or action | Shortcut | What it does | | ------------------------------ | ------------------------- | ----------------------------------------------------------------------------- | | **Select** | `A` | Drag a clip to move it, drag an edge to trim, double-click to edit properties | | **Blade** | `B` | Click a clip to cut it at that point | | **Trim** | `T` | Drag an edge to ripple; drag inside a clip to slip its footage | | **Split at playhead** | `S` | Cuts the selected clips at the playhead | | **Delete** / **Ripple delete** | `Delete` / `Shift+Delete` | Removes clips; ripple delete also closes the space | | **Duplicate** | `Cmd/Ctrl+D` | Copies the selection | | **Disable clip** | `D` | Turns a clip off without removing it | | **Close gap** | none | Removes an empty gap between clips (select the gap marker) | Right-click a clip for more: **Detach audio** or **Link audio**, **Replace with selected media**, **Update to new version**, **Reload from source**, **Turn off effects**, **Reset effects**, **Edit properties** and **Set range to selection**. **Snap** (**Snapping on** / **Snapping off**) aligns edits to clip edges and the playhead. **Move audio with video** keeps a video clip and its linked audio together. Edits always land on whole frames of the sequence frame rate. ## Set the sequence size, frame rate and length Select the sequence summary in the top bar, or deselect everything and open **Properties**. | Setting | Options | | ------------------- | ----------------------------------------------------------------------------------------------------------- | | **Preset** | **1080p** 1920×1080, **720p** 1280×720, **4K** 3840×2160, **9:16** 1080×1920, **1:1** 1080×1080, **Custom** | | **Size** | Width and height, 16 to 16,384 px | | **Frame rate** | 24, 25, 30, 50 or 60 fps, or **Custom** (1 to 240) | | **Length** | Follows the clips: the sequence ends where the last clip ends | | **Default framing** | **Fit** or **Fill** for clips whose shape differs from the sequence | ## Adjust clips and tracks Select a clip and open **Properties**. Video clips have **Framing** (**Fit**, **Fill**, **Custom**), **Opacity**, **Position**, **Scale**, **Rotation**, **Anchor**, crop, and a composite **Mode** (**Normal**, **Multiply**, **Screen**, **Overlay**, **Add**). Audio clips have **Gain**, **Fade In** and **Fade Out**. To add a track, right-click an empty part of the timeline and choose **Add video track** or **Add audio track**, or drop a clip into the new-track area. Track headers let you rename a track (double-click), change its color, lock it, mute or solo it, turn it on or off, and delete it. Select a track header to edit effects for every clip on that track. ## Mix audio Select **Mixer** in the toolbar. Each audio track gets a strip with level, pan, mute and solo, plus **Equalizer** and **Dynamics** (a compressor followed by a limiter). The **Master** strip controls the whole mix. ## Preview playback | Shortcut | Action | | ---------------------------- | --------------------------------------------------------------------- | | `Space` | Play / pause | | `J` / `K` / `L` | Reverse / stop / forward; hold `K` and tap `J` or `L` to step a frame | | `Left` / `Right` | Back or forward one frame | | `Shift+Left` / `Shift+Right` | Back or forward one second | | `Up` / `Down` | Previous or next cut | | `Home` / `End` | Go to start or end | | `I` / `O` / `Alt+X` | Set range start / set range end / clear range | | `F` / `Shift+F` | Fit timeline to selection or range / fit track height | | `?` | Show all Timeline shortcuts | Shortcuts work while the timeline has focus. The loop button cycles **Loop range**, **Loop all** and **No loop**. The preview chip reports **Preview ready**, preparation progress or dropped frames, and sets preview **Quality** to **Full**, **Half**, **Quarter** or **Eighth**. ## Render the Timeline into the graph 1. Select **Render** in the top bar. 2. Choose a preset: **Custom**, **Timeline** (current size and fps), **Preview** (720p WEBM), **MP4**, **High** or **Social** (1080p MP4). 3. Set **Range** to **Whole timeline** or **Range**, and the file **Name**. 4. Open **More settings** for **Container** (WEBM or MP4), bitrate, resolution, frame rate and audio options. 5. Select **Render**. Cancel at any time from the progress window. The finished render feeds the node's **Video** output. The **Render** popover then shows **Up to date** or **Out of date**, and offers **Download**. Details of every setting are in [Export finished media](https://docs.nodespell.com/edit/export/). ## Frequently asked questions ### Why is MP4 unavailable when I render a Timeline? Nodespell renders in your browser, and MP4 needs H.264 and AAC encoding support. When the browser lacks it, MP4 shows as unsupported and WEBM is used. ### Does the Timeline output update automatically when I edit? No. The **Video** output keeps the last render until you select **Render again**. The popover marks it **Out of date** when the edit has changed. ### Can I put a still image on the Timeline? Yes. Image clips go on video tracks and start five seconds long; trim them like any other clip. ### Can I put audio clips on a video track? No. Tracks are type-locked: audio clips stay on audio tracks. ## Related - [Video editing outside the Timeline](https://docs.nodespell.com/edit/video-editing/): single-purpose video nodes such as Trim Video and Join Video - [Export finished media](https://docs.nodespell.com/edit/export/): render settings and downloads - [Connections](https://docs.nodespell.com/graph/connections/): wire the rendered video onward - [Takes](https://docs.nodespell.com/graph/takes/): how new takes reach the Timeline - [Add a soundtrack to a clip](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/): a worked Timeline example --- # Edit video with single-purpose video nodes > Trim, join, resize, retime, reverse, transform, transcode and grab frames from video in Nodespell with video nodes, and know when to use the Timeline. Source: https://docs.nodespell.com/edit/video-editing/ · Last updated: 2026-11-10 Video nodes in Nodespell are single-purpose graph nodes that change one video in one way: trim it, join clips, resize, retime, reverse, transform, transcode, or extract a still frame. Each node takes a video input, runs in your browser, and outputs a new video or image, so the edit stays in your graph and you can run it again on a new take. ## Which video node do I need? All of these nodes are in the [Tab menu](https://docs.nodespell.com/graph/add-nodes/) under their names. Each one leaves the source video unchanged. | Node | What it does | Output | | ------------------- | ------------------------------------------------------------------------------------- | -------------------------- | | **Trim Video** | Cuts from the start and end, or reduces the clip to a target length | MP4 video | | **Join Video** | Joins connected videos end to end in connection order | MP4 video | | **Resize Video** | Changes the frame size by preset, scale, custom size or a reference input | MP4 video | | **Retime Video** | Changes playback speed, or fits the clip to a target duration | MP4 video | | **Reverse Video** | Plays the clip backwards | MP4 video | | **Transform Video** | Scales, moves, rotates, skews or mirrors the picture inside its frame | MP4 video | | **Transcode Video** | Re-encodes to MP4, MOV, WebM or MKV with quality, size, frame-rate and audio controls | Video in the chosen format | | **Extract Frame** | Takes one still from the video | Image | Upscaling a video is an AI model task rather than a video node; see [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/). ## Add a video tool from a viewer or node You do not have to search the Tab menu. Open a video result in the full-screen viewer, or right-click a node that outputs video. The video tools there are **Trim**, **Extract frame**, and, under **More tools** (**Add as a node**), **Transcode Video**, **Resize**, **Speed**, **Reverse** and **Upscale**. Choosing one adds the matching node after the current one and connects it. ## Trim a video 1. Add a **Trim Video** node after the video, or choose **Trim** in the full-screen video viewer. 2. Drag the start and end handles to choose what to keep. The readout shows **Keeps** with the kept length and frame count. 3. In the full-screen viewer, select **Create Trim Video** to add the connected node. On an existing node, changes save as you drag. To trim an existing **Trim Video** node visually, right-click it and choose **Edit Trim**. The trimmer has these keys: | Key | Action | | ---------------- | ----------------------------------------- | | `Space` | Play or pause | | `I` / `O` | Set the In or Out point to the playhead | | `Left` / `Right` | Step one frame (`Shift` steps one second) | | `Home` / `End` | Jump to the start or end | **Trim Video** parameters: | Parameter | Options | | --------------------------------------- | -------------------------------------------------------------------------------------------------------------------- | | **Trim mode** | **Start and end** (default) or **Target length** | | **Unit** | **Seconds** or **Frames** | | **Trim from start** / **Trim from end** | Amount to remove at each end | | **Target source** | **Enter length**, or **Reference input** to match the duration of a connected video or audio on **Length reference** | | **Trim direction** | **Trim end**, **Trim start** or **Trim symmetrically** when reducing to a target length | ## Join, resize and retime - **Join Video** takes several videos on its **Videos** input in connection order. **Resolution** can **Use largest** (default), **Use smallest**, **Use most common**, **Use first input**, **Use last input**, or **Custom**. **Fit mode** is **Fit**, **Fill** or **Stretch**. **Export audio** and **Audio gain** control the joined sound. - **Resize Video** sets the size by **Size preset** (4K, HD, 720p and square, in landscape or portrait), **Scale factor** (0.05 to 8), **Custom size** (2 to 8,192 px), or **Reference**, which copies the size of a connected image or video. - **Retime Video** sets **Speed** (0.1 to 10) or a target **Duration** in seconds. **Retime audio** keeps the sound in step. - **Reverse Video** has one option, **Reverse audio**, which is off by default. ## Transcode a video for delivery **Transcode Video** re-encodes on your device. Choose **Format** (MP4, MOV, WebM, MKV), **Rate control** (**Quality**, **Bitrate** or **Approximate file size**), **Video codec** (available codecs depend on your browser), **Maximum size**, **Maximum frame rate**, and audio options. It never enlarges the picture or raises the frame rate. It accepts SDR video up to 256 MiB and does not keep transparency. ## Grab a still frame **Extract Frame** outputs one image. Choose **First Frame**, **Last Frame**, **By Frame Number**, **By Timestamp** or **By Percentage**. Frame numbers and timestamps accept negative values to count from the end. In the full-screen video viewer, **Create Extract Frame node** grabs the frame you are on. ## Should I use the Timeline or a video node? | Use a video node when | Use the [Timeline](https://docs.nodespell.com/edit/timeline/) when | | --------------------------------------------------------------- | ------------------------------------------------------------- | | One clip needs one change: trim, resize, speed, reverse, format | You are cutting several clips together with precise placement | | The same edit should apply to each new take you run | You need separate audio tracks, fades, mixing or EQ | | You are joining clips back to back with no overlap | You need overlaps, per-clip framing, opacity or blend modes | | You are preparing a clip for another model's input | You want to scrub and review the cut while editing | You can combine both: trim or retime clips with nodes, then connect them to a Timeline to assemble the cut. ## Frequently asked questions ### Do video nodes use credits? The video nodes on this page process video in your browser rather than running an AI model. Upscaling is a model task and shows its own estimate. ### Why does Transcode Video offer fewer codecs than I expect? The codecs come from your browser's encoder support. WebM needs VP8, VP9 or AV1; choose **Auto** if a combination is not available. ### What order does Join Video use? Join Video uses the order of the connections to its **Videos** input. ### Can I trim by frames instead of seconds? Yes. Set **Unit** to **Frames** on the Trim Video node, or set **Show times as** to **Frames** in the trimmer. ## Related - [Timeline](https://docs.nodespell.com/edit/timeline/): multi-track editing, mixing and rendering - [Export finished media](https://docs.nodespell.com/edit/export/): download and delivery formats - [Viewers](https://docs.nodespell.com/edit/viewers/): the full-screen video viewer and its tools - [Connections](https://docs.nodespell.com/graph/connections/): chaining nodes and multi-inputs - [Supported media](https://docs.nodespell.com/reference/supported-media/): which video files you can upload --- # Download and export finished media > Get finished images, video and audio out of Nodespell by downloading from nodes, viewers and the Asset Manager, or rendering a Timeline to MP4 or WebM. Source: https://docs.nodespell.com/edit/export/ · Last updated: 2026-11-10 Exporting from Nodespell means saving a result file to your computer. You download a node's result, a file in a viewer, or an Asset from the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/), and Nodespell saves the stored file at its full resolution, not a preview. To produce a new delivery file, such as an edit, a different container or a smaller size, you render a [Timeline](https://docs.nodespell.com/edit/timeline/) or run a [video node](https://docs.nodespell.com/edit/video-editing/) first. ## Where can I download from? | Where | Control | What you get | | ------------------------ | -------------------------------------------------------------------------- | ---------------------------------------- | | A node on the graph | Right-click → **Download**, or **Download N assets** for several | The current result of each selected node | | The node's preview | **Download** button on the preview toolbar | The result shown on the node | | Full-screen image viewer | **Save image** | The image you are viewing | | Full-screen video viewer | **Download video** | The video you are viewing | | Full-screen audio viewer | **Download audio** | The audio you are viewing | | Text viewer | **Download text** | The text result | | 3D viewer | **Download Model** | The model file | | Asset Manager | Right-click → **Download Asset**, or **Download N assets** for a selection | The original uploaded or generated file | | Timeline editor | **Render** → **Download** | The last Timeline render | When you download several files at once, Nodespell starts one download per file. It does not bundle them into a ZIP. If some fail, a notice says how many could not be downloaded. ## Download a result from a node 1. Select one or more nodes in Graph. Choose the [take](https://docs.nodespell.com/graph/takes/) you want first if a node has several. 2. Right-click a selected node and choose **Download**. 3. Your browser saves each file. Nodespell names files from the Project, the node and the time the result was made. Prompt nodes have no file to download. Asset nodes download the file they point to. ## Does downloading keep full resolution? Yes. Downloads use the stored result or the original Asset file, not the viewer's preview or thumbnail. Lowering preview quality in a viewer or in the Timeline changes only playback, not the downloaded file. An image result keeps its stored format, for example PNG or, for HDR results, EXR. ## Export an edited video from the Timeline The Timeline is where you set export options for video. Select **Render** in the Timeline top bar. The render settings are: | Setting | Options | Default | | ------------------- | -------------------------------------------------------------------------------------------- | ------------------- | | **Preset** | **Custom**, **Timeline**, **Preview** (720p WEBM), **MP4**, **High**, **Social** (1080p MP4) | **Custom** | | **Range** | **Whole timeline** or **Range** (In/Out points) | **Whole timeline** | | **Name** | File name for the download | `timeline-export` | | **Container** | **WEBM** or **MP4** | **WEBM** | | **Bitrate** (video) | 500 to 50,000 kb/s | 8,000 kb/s | | **Resolution** | **Timeline resolution** or **Custom** (16 to 16,384 px) | Timeline resolution | | **Frame rate** | Timeline rate or **Custom** (1 to 120 fps) | Timeline rate | | **Include audio** | On (**Timeline audio mix**) or off (**Video only**) | On | | **Codec** (audio) | **Auto for container**, **AAC**, **Opus** | Auto | | **Bitrate** (audio) | 16 to 512 kb/s | 128 kb/s | | **Channels** | **Stereo** or **Mono** | Stereo | Track mute, solo, gain, clip gain, fades and track effects are included in the mix. MP4 is available only when your browser can encode H.264 and AAC; otherwise it shows **MP4 (unsupported)**. After rendering, select **Download** in the same popover. The render is made and stored in your browser on this device, and it also feeds the Timeline node's **Video** output. > **Caution:** A Timeline render is stored on this device until you download it or a later step uploads it. Download the file if you need a copy before you clear browser data or switch computers. ## Change format, size or frame rate without the Timeline For a single clip, add a node instead of opening the Timeline: - **Transcode Video** writes MP4, MOV, WebM or MKV, with quality, bitrate or approximate file-size control, a maximum size and maximum frame rate, and audio options. - **Resize Video** outputs an MP4 at a new frame size. - **Retime Video** and **Trim Video** change duration. Run the node, then download its result. See [Video editing](https://docs.nodespell.com/edit/video-editing/) for every option. ## Export from a compositor In a compositor's **Colour** settings, **Result** chooses what it produces: **Review · as viewed (PNG)**, **Review · as viewed (video)**, **Master · linear EXR**, or **HDR review · 10-bit WebM PQ** where the device supports it. **Export EXR sequence…** writes frames to a folder you choose and needs desktop Chrome or Edge. See [Color management](https://docs.nodespell.com/edit/color-management/). ## Export a 3D Asset In the Asset Manager, right-click a supported 3D Asset and choose **Export 3D Model** (or **Export DCC Format**), then **Export Alembic (.abc)**. ## Frequently asked questions ### Can I export a whole Project as a video? No. Nodespell exports results, not Projects. Assemble the shots you want in a [Timeline](https://docs.nodespell.com/edit/timeline/) and render it. ### Why is my Timeline download a WEBM file? WEBM is the default container. Choose the **MP4** or **Social** preset, or set **Container** to MP4, if your browser supports MP4 encoding. ### Can I download several results as one ZIP? No. Multi-select downloads start one download per file. ### Are downloaded files resized? No. Downloads are the stored files themselves; Nodespell does not resize or re-encode them on download. ## Related - [Timeline](https://docs.nodespell.com/edit/timeline/): build the edit you export - [Video editing](https://docs.nodespell.com/edit/video-editing/): Transcode, Resize and other video nodes - [Preview, download and Trash](https://docs.nodespell.com/assets/preview-download-and-trash/): downloading from the Asset Manager - [Supported media](https://docs.nodespell.com/reference/supported-media/): output formats at a glance - [Takes](https://docs.nodespell.com/graph/takes/): choose which result a node outputs --- # Get started with Agent > Use Nodespell Agent, the in-product assistant, to explain your graph, make edits, and prepare generations that run only after you approve them. Source: https://docs.nodespell.com/agent/ · Last updated: 2026-11-10 Agent is the AI assistant built into Nodespell. In a conversation, Agent can explain what a graph does, find and change nodes, connect them, and prepare a generation. Agent never spends credits on a generation without your approval. In the app, Agent lives in the **Assistant** panel. ## Open the Assistant panel Agent works with the Project that is open. It reads and changes the graph that the Graph Editor shows, so keep a graph visible while you work with it. The table lists every way to open the panel. | Where you are | How to open Agent | | --------------------------- | ------------------------------------------------------------------------------------ | | **Graph** mode | Open the add-panel menu in any panel area and choose **Automation** > **Assistant**. | | Any mode, from the menu bar | **Window** > **New Panel Window** > **Assistant** opens Agent as a floating window. | | Command palette | Run **Add Assistant Panel**. | | **Create** mode | Select **Assistant** in the tool rail. | | **Immersive** mode | Select **Assistant** in the tool rail. | If Agent reports **Open a graph panel first.**, open a Graph Editor and send the request again. To learn more about the modes, see [Workspace modes](https://docs.nodespell.com/start/workspace-modes/). For panels, see the [interface tour](https://docs.nodespell.com/start/interface/). ## Walk through a first request A new conversation opens with **What would you like to create?** and a few suggested requests under **Try**. The suggestions change to match what you have selected. For example, with an image node selected, one suggestion is **Turn the selected image into a video**. 1. Select one or more nodes on the canvas if your request is about them. Then select the **Include selected nodes** chip above the message box. Agent only knows about the nodes you include. See [Context and references](https://docs.nodespell.com/agent/context-and-references/). 2. Check the mode menu next to the **+** button. **Edit** is the default. It lets Agent make changes to your graph. To ask questions without making changes, choose **Ask**. 3. Type a request in the message box, for example "Add an upscaler after this image and connect it." Press `Enter` to send it. Press `Shift+Enter` to start a new line. 4. Watch the progress steps while Agent reads the graph and works. When the turn ends, Agent summarizes what it did. Select a referenced node chip to move the canvas to that node. 5. If Agent shows a **Needs your approval** card, read what it wants to do, then select **Approve** or **Deny**. Deleting nodes and starting a paid generation always ask first. 6. Review the graph. If you don't want a change Agent made, undo it with `Cmd/Ctrl+Z` like any other graph edit. ## What can Agent do, and what can't it do? | Agent can | Agent can't | | --------------------------------------------------------------------- | ----------------------------------------------------------------------- | | Explain a graph, a node, or how a branch is connected | Change the graph in **Ask** or **Plan** mode | | Find nodes in your Project and in the node and Creative Tool catalogs | Spend credits on a generation without your approval | | Create, duplicate, rename, configure, connect, and disconnect nodes | Delete nodes or switch a Creative Tool's model without your approval | | Create a Creative Tool and switch its model | Edit inside the Timeline or a compositor, or arrange the layout for you | | Find Project Assets and place them on the graph | Use Snippets or Custom Nodes, or import from the web | | Estimate, prepare, start, track, and cancel a run | Work on a different Level while a request is running | For the complete list, with an example request for each action, see [What Agent can do](https://docs.nodespell.com/agent/what-agent-can-do/). ## Does Agent cost credits? Yes. Every Agent request uses credits for Agent's own reasoning. The cost depends on the Agent model you choose and on how much work the request takes. A generation that Agent starts uses credits the same way a run you start yourself does. Both kinds of charge appear in the **Usage** tab of the Tasks panel. See [Sessions and models](https://docs.nodespell.com/agent/sessions-and-models/). ## Frequently asked questions ### Is Agent the same as the Assistant panel? Yes. Agent is Nodespell's name for the assistant, and **Assistant** is the label on the panel and tool rail button that opens it. ### Does Agent see my whole graph? No. Agent gets the nodes you include in a message. When it needs more, Agent looks up other parts of the Project, and the steps it takes appear in the conversation. ### Can Agent run a generation by itself? No. Agent can estimate and prepare a run, but it must ask for your approval before the run starts. You see an approval card, and nothing runs until you select **Approve**. ### Can I undo something Agent did? Yes. You can undo Agent's graph edits with `Cmd/Ctrl+Z`. Undo does not reverse a generation that has already started. You can ask Agent to cancel a run, or cancel it in the Tasks panel. ## Related - [Modes and approvals](https://docs.nodespell.com/agent/behavior-modes-and-approvals/): Ask, Plan, and Edit, and which actions always need your approval. - [Context and references](https://docs.nodespell.com/agent/context-and-references/): Include nodes, mention them with @, and attach reference files. - [Generate with Agent](https://docs.nodespell.com/agent/generate-with-agent/): Estimates, run approvals, and the Results tab. - [Sessions and models](https://docs.nodespell.com/agent/sessions-and-models/): Conversations, the Agent model, credits, and Stop, Steer, and Resume. - [What Agent can do](https://docs.nodespell.com/agent/what-agent-can-do/): Every capability, with example requests and limits. --- # Choose an Agent mode and approve actions > Learn how Nodespell Agent's Ask, Plan, and Edit modes control graph changes, and which actions always wait for you to approve or deny them. Source: https://docs.nodespell.com/agent/behavior-modes-and-approvals/ · Last updated: 2026-11-10 Agent modes in Nodespell control what Agent may do to your graph during a request. **Ask** only answers. **Plan** proposes changes without making them. **Edit** makes the changes. In every mode, Agent still stops and asks before higher-impact actions, such as deleting nodes or starting a paid generation. ## What does each Agent mode do? The mode menu sits next to the **+** button below the message box. Each option shows a short summary. Hover over an option to see its full description. | Mode | Summary in the menu | What Agent may do | | -------- | ------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------- | | **Ask** | Answers only | Answers questions about your project. Never changes it. Agent can read the graph, search catalogs and Assets, look at results, and estimate costs. | | **Plan** | Proposes first | Reads your project and proposes changes. The graph does not change. | | **Edit** | Edits directly | Makes graph edits during the request. Still asks before spending credits, and before the other actions listed below. | **Edit** is the default. Nodespell remembers the mode you choose in this browser. The mode applies to each message when you send it. A follow-up that you queue keeps the mode that was selected when you queued it. You can't change the mode while Agent is working. ## Change the mode before a request 1. Open the mode menu. It shows the current mode, for example **Edit**. 2. Choose **Ask**, **Plan**, or **Edit**. 3. Type your request and send it. If Agent tries to change the graph in **Ask** or **Plan**, Nodespell refuses the change and Agent reports that it can't make it. To carry out a plan, switch to **Edit** and ask Agent to apply the plan it proposed. ## Which actions always need your approval? In **Edit**, Agent makes reversible graph edits on its own. The actions in the table below always pause for your decision, even in **Edit**. | Action | Applies without asking in Edit | Needs approval | | ---------------------------------------------------- | ------------------------------ | -------------- | | Create, duplicate, or rename nodes | Yes | No | | Change parameter values | Yes | No | | Connect or disconnect nodes | Yes | No | | Create a Creative Tool, or place Assets on the graph | Yes | No | | Delete nodes | No | Yes | | Switch a Creative Tool's model | No | Yes | | Prepare a run (a paid generation) | No | Yes | | Start a prepared run | No | Yes | | Cancel a run | No | Yes | Reading, searching, inspecting results, and estimating costs never change anything, so they never ask. ## Approve or deny an action When Agent reaches one of these actions, the conversation shows a **Needs your approval** card that names the step. Agent waits until you decide. - Select **Approve** to let Agent continue with that exact step. - Select **Deny** to refuse the step. Agent continues without it, or explains what it couldn't do. After you decide, the card becomes a single line: **Approved**, **Denied**, or **Expired**. Each approval covers one action, based on the graph as it was when Agent asked. If the graph changes before that action runs, Nodespell rejects the action. Agent has to check the graph again and ask again. For generation approvals, see [Generate with Agent](https://docs.nodespell.com/agent/generate-with-agent/). Sometimes Agent needs more information instead of an approval. It then shows a **Needs your input** card with up to six choices. Select one of the choices, or type your own answer in the message box. ## How do you undo what Agent changed? Each graph edit Agent makes goes into your normal undo history. Press `Cmd/Ctrl+Z` to step back through Agent's changes, the same way you undo your own edits. Selecting **Stop** keeps any changes Agent already made. Undo them if you don't want them. > **Caution:** Undo does not reverse a generation that has already started. Credits are used once a run is submitted. Ask Agent to cancel the run, or cancel it from the [Tasks panel](https://docs.nodespell.com/graph/tasks/). ## Frequently asked questions ### What is the default Agent mode? **Edit** is the default. After you choose a different mode, Nodespell remembers it in the same browser. ### Can Agent change my graph in Plan mode? No. In **Plan**, Agent reads the graph and describes the changes it would make. Switch to **Edit** and ask Agent to apply the plan. ### Does Edit mode let Agent spend credits without asking? No. In **Edit**, Agent asks for approval before it prepares a run and again before it starts the run. Agent's own reasoning still uses credits for each request. ### Why did my approval stop working? An approval covers one action, based on the graph as it was when Agent asked. If you or Agent change the graph first, Nodespell rejects that approval. Agent has to check the current graph and ask again. ## Related - [Agent overview](https://docs.nodespell.com/agent/) — open Agent and send a first request. - [Generate with Agent](https://docs.nodespell.com/agent/generate-with-agent/) — estimates and run approvals step by step. - [What Agent can do](https://docs.nodespell.com/agent/what-agent-can-do/) — every action and whether it changes the graph. - [Run nodes](https://docs.nodespell.com/graph/run-nodes/) — how runs work when you start them yourself. - [Tasks panel](https://docs.nodespell.com/graph/tasks/) — watch and cancel running work. --- # Give Agent context with nodes and files > Point Nodespell Agent at the right nodes with Include selected nodes and @ mentions, and attach reference files with roles such as Character or Style. Source: https://docs.nodespell.com/agent/context-and-references/ · Last updated: 2026-11-10 Context is what Nodespell Agent knows about a request before it starts working. You provide context by including selected nodes, mentioning nodes with `@`, and attaching reference files. Clear context makes Agent faster and more accurate, because it doesn't have to search the Project to guess which nodes you mean. ## What context does Agent get by default? Agent does not automatically receive everything you have selected on the canvas. Each message carries only these things: | Context | How it gets into a message | | ----------------------- | ------------------------------------------------------------------- | | Nodes you include | **Include selected nodes**, or a node chip you leave in the message | | Nodes you mention | `@` followed by a node name | | Files you attach | **Attach files**, drag and drop, or paste | | The conversation so far | Earlier messages in the same conversation | | Where you are | The current Level and the size of the graph | When Agent needs more, it uses its own read-only lookups. Agent can search the current Level or the whole Project, inspect nodes, and follow connections upstream or downstream. Each lookup appears as a step in the conversation, for example **Searched the project for "portrait"**. For Levels, see [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/). A follow-up that points back with words such as "it", "them", or "the same", for example "make it brighter", reuses the nodes from your previous request. This happens only when the follow-up doesn't include nodes of its own. ## Include selected nodes in a message 1. Select nodes on the canvas. A chip appears above the message box. With one node selected, it reads **Include** followed by the node's name. With several, it reads **Include 3 selected nodes**. 2. Select the chip, or open **+** and choose **Include selected nodes**. Each node becomes a removable chip with a thumbnail. 3. Remove any chip you don't want by selecting its **×**. 4. Type your request and send it. One message can include up to 60 nodes. If you select more, only the first 60 are included. The tooltip warns you before you add them. After three chips, the remaining nodes collapse into a **+N more** chip that you can expand. When you pick one of the suggested requests in a new conversation while nodes are selected, Agent includes the selection for you. ## Mention a node with @ Type `@` in the message box, or open **+** and choose **Mention a node**. A list of nodes on the current Level appears. Each row shows the node's name, its type, and a thumbnail. Keep typing to filter the list, which shows up to eight matches. Press `Enter` or `Tab` to insert the highlighted node, or press `Escape` to close the list. Mentions work well when the node isn't selected, or when several nodes have similar names. ## Attach reference files Open **+** and choose **Attach files**, drag files onto the message box, or paste them. Agent uploads each file to the Project's Assets in the **Uploads** folder. Agent then passes the file to the request as an Asset reference. Agent can place that Asset on the graph and connect it. For the Asset library, see [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). | Limit | Value | | ----------------- | -------------------------------------------------------------------------------------------- | | Files per message | Up to 12 | | Size per file | 200 MB or smaller | | Supported types | The same types you can upload to Assets. See [Supported media](https://docs.nodespell.com/reference/supported-media/). | Each attachment chip shows its status: **Queued**, upload progress as a percentage, **Processing**, **Ready**, or **Failed**. You can send the message only after every attachment is ready. If an upload fails, select the retry button on the chip or remove the file. If you send attachments without any text, Agent receives **Use the attached references.** as the request. ### Choose a reference role Each attachment chip has a role menu that tells Agent how to use the file. New attachments start as **Source**. | Role | Use it for | | --------------- | ------------------------------------------ | | **Source** | The material to work on directly | | **Character** | A person or character to keep consistent | | **Style** | A look, palette, or rendering style | | **Environment** | A location or setting | | **Motion** | Movement or camera reference | | **Audio** | Sound, voice, or music reference | | **Other** | Anything else; describe it in your message | The role is a hint. Agent decides how to wire the file into the graph, and you can see the result on the canvas. ## Frequently asked questions ### Does selecting nodes on the canvas send them to Agent? No. Selecting nodes shows an **Include selected nodes** chip, but the nodes reach Agent only when you include them or mention them with `@`. ### How many files can I attach to an Agent message? You can attach up to 12 files at a time, and each file must be 200 MB or smaller. ### Where do attached files go? Agent uploads attached files to the Project's Assets in the **Uploads** folder. They stay there after the conversation ends, like any other uploaded Asset. ### Can Agent find nodes I didn't mention? Yes. Agent can search the current Level or the whole Project by name and content, and it shows each search as a step. Naming the node yourself is faster and avoids mistakes when names are similar. ## Related - [Agent overview](https://docs.nodespell.com/agent/) — open Agent and send a first request. - [Modes and approvals](https://docs.nodespell.com/agent/behavior-modes-and-approvals/) — what Agent may do with that context. - [What Agent can do](https://docs.nodespell.com/agent/what-agent-can-do/) — request examples that use mentions and attachments. - [Upload and organize Assets](https://docs.nodespell.com/assets/upload-and-organize/) — manage the files Agent uploads. - [Supported media](https://docs.nodespell.com/reference/supported-media/) — file types you can attach. --- # Generate images and video with Agent > Ask Nodespell Agent to estimate, prepare, and start a generation after your approval, then review, compare, choose, and revise takes in the Results tab. Source: https://docs.nodespell.com/agent/generate-with-agent/ · Last updated: 2026-11-10 Generating with Agent in Nodespell means asking Agent to run part of your graph for you. Agent estimates the credit cost, prepares the exact run, and asks for your approval. After you approve, Agent starts the run, follows its progress, and collects the finished takes in the **Results** tab. There you can review, compare, choose, and revise them. ## How does a generation with Agent work? A generation with Agent always passes through the same four stages. Only the last two can spend credits, and each of those waits for you. | Stage | What Agent does | Spends credits | Needs your approval | | -------- | -------------------------------------------------------------------------------------------------------------------- | -------------- | ------------------- | | Estimate | Calculates the expected credit cost of running the target nodes | No | No | | Prepare | Checks the run plan, your balance, model access, and a credit cap, then reserves an authorization for that exact run | No | Yes | | Start | Submits the prepared run | Yes | Yes | | Follow | Tracks progress and reports the takes as they arrive | No | No | Agent's own reasoning uses credits for every request, separately from the generation. See [Sessions and models](https://docs.nodespell.com/agent/sessions-and-models/). ## Run a generation with Agent 1. Make sure the mode menu shows **Edit**. In **Ask**, Agent can estimate the cost but can't prepare or start a run. 2. Include the node you want to run, or mention it with `@`. Then describe the run, for example "Run @Hero shot twice and show me both takes." See [Context and references](https://docs.nodespell.com/agent/context-and-references/). 3. Read the estimate in Agent's reply. If you're unsure, ask "What will this cost?" before you approve anything. A prepared run can produce one to four outputs. 4. When the **Needs your approval** card appears for preparing the run, select **Approve**. Select **Deny** to stop here without spending generation credits. 5. When the second **Needs your approval** card appears for starting the run, select **Approve** to submit it. 6. Watch the run card. It moves through **Queued** and **Running** to **Completed**, **Partially completed**, **Failed**, or **Cancelled**. It also lists progress and the takes it produced. If your graph uses nodes whose cost can only be known after they run, the estimate is a lower bound. Agent must then set an explicit credit cap before it can ask for approval. Credit estimates are included in paid plans. On other plans, Agent tells you that an estimate isn't available instead of giving a number. For plans, see [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Why does Agent need to prepare a run again? A prepared run is tied to the graph exactly as it was when Agent prepared it. Nodespell refuses to start the run and Agent prepares it again when any of these happen: | Situation | What happens | | ---------------------------------------------- | ----------------------------------------------------- | | You or Agent changed the graph after preparing | The run is refused. Agent must prepare and ask again. | | The estimate rose above the approved cap | The run is refused. Agent must prepare and ask again. | | More than five minutes passed since preparing | The authorization expires. Agent must prepare again. | | Your balance or model access changed | Nodespell rejects the run. Nothing is submitted. | To refresh an out-of-date estimate, ask Agent to estimate or prepare the run again on the current graph. ## Review takes in the Results tab When the conversation references media takes, the top of the panel shows two tabs: **Chat** and **Results**, with a count. **Results** opens **Session results**, a gallery of the takes from this conversation. It includes takes from runs Agent started and results Agent inspected. The gallery shows takes from nodes on the current Level. | Control | What it does | | ------------------------ | --------------------------------------------------------------------------------------------------------------- | | **Review** | Opens the take in the full-screen viewer for its media type | | **Open result** | Opens a take that has no inline preview | | **Choose take** | Makes this take the node's selected output, so downstream nodes use it | | **Revise** | Returns to **Chat** with a follow-up started for you: keep the approved work and change only what you type next | | Focus node button | Moves the canvas to the node that produced the take | | **Compare** (video only) | Marks a video for comparison | | **Compare selected** | Appears when two videos are marked, and opens them side by side | The gallery shows the 24 most recent takes. A note tells you how many older results remain in the session history. For take selection on the canvas, see [Takes](https://docs.nodespell.com/graph/takes/). For the viewer, see [Viewers](https://docs.nodespell.com/edit/viewers/). ### What is a usable partial result? A take labeled **Usable partial result** came from a run that finished only part of its work. You can review it and choose it like any other take. Ask Agent to check the run if you need the rest. > **Note:** Stopping Agent does not always stop a submitted run. Completed changes are kept, and work already submitted may still finish and use credits. Ask Agent to cancel the run, which needs your approval, or cancel it from the [Tasks panel](https://docs.nodespell.com/graph/tasks/). ## Frequently asked questions ### Will Agent start a paid generation without asking? No. Agent asks for approval before it prepares a run and again before it starts the run, in every mode. ### How many takes can Agent generate in one run? A prepared run produces between one and four outputs. Ask for the number you want, for example "make three takes." ### Why was my approved run not started? The graph probably changed, the estimate rose above the approved cap, or the five-minute authorization expired. Ask Agent to prepare the run again on the current graph. ### Does Choose take change which result the graph uses? Yes. **Choose take** makes that take the node's selected output, the same as choosing it on the canvas. ## Related - [Modes and approvals](https://docs.nodespell.com/agent/behavior-modes-and-approvals/) — which actions always ask first. - [Run nodes](https://docs.nodespell.com/graph/run-nodes/) — runs, reuse of earlier results, and estimates outside Agent. - [Takes](https://docs.nodespell.com/graph/takes/) — browse and choose results on the node. - [Tasks panel](https://docs.nodespell.com/graph/tasks/) — progress, failures, and cancelling. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) — what pays for generations. --- # Manage Agent conversations, models, and credits > Manage Nodespell Agent conversations, choose the Agent model and reasoning effort, understand Agent credit use, and stop, steer, resume, or queue requests. Source: https://docs.nodespell.com/agent/sessions-and-models/ · Last updated: 2026-11-10 An Agent session in Nodespell is a conversation. It keeps your requests, Agent's replies, approvals, and results together for one Project. You choose which AI model Agent thinks with, and that choice affects speed, quality, and credit use. While Agent works, you can stop it, redirect it, or queue the next request. ## Manage conversations The header of the **Assistant** panel shows the current conversation's title. A new, untitled conversation is called **New chat**. Agent titles a conversation from your first message. | Task | How | | -------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------- | | Start a conversation | Select the **+** button (**New chat**) in the header. | | Switch conversations | Select the title to open the list, then pick a conversation. The list is grouped into **Today**, **Last 7 days**, and **Earlier**. | | Search | Type in **Search conversations** at the top of the list. The search matches titles, including archived conversations. | | Rename | Open **Conversation options** (**…**) and choose **Rename**. Type a name of up to 48 characters, then press `Enter` to save or `Escape` to cancel. | | Archive | Open **Conversation options** and choose **Archive**. The conversation moves to the **Archived** submenu at the bottom of the list. | | Restore | Open the **Archived** submenu and select the conversation. | | Delete | Open **Conversation options**, choose **Delete conversation**, and confirm. | Archiving only hides a conversation. It doesn't change the graph or its takes. You can't switch, rename, archive, or delete conversations while Agent is working. > **Caution:** **Delete conversation** removes the conversation and its complete history, and this cannot be undone. Graph changes and takes that Agent produced stay in the Project. ## Choose the Agent model and reasoning effort The model menu sits at the right of the message box. It shows the current model's name. The menu groups models by cost, cheapest first. Hover over a model to see what it is for and its ratings for **Speed**, **Accuracy**, and **Cost**. | Group | Typical use | | ------------------ | ------------------------------------------------------------------------------- | | **Cost effective** | Quick, low-cost replies for simple questions and small edits | | **Regular** | Everyday graph building and planning. One model here is marked **Recommended**. | | **High quality** | Long, multi-step work and hard problems. Slower, and costs more. | Agent starts with a model from the **Cost effective** group. Nodespell remembers your choice in this browser. The list of models can change over time, so rely on the menu rather than a fixed list. When the selected model supports it, the menu also shows **Reasoning effort**, with **Low**, **Medium**, **High**, and **Extra high**. **Medium** is the default. Higher effort can take longer and use more credits. The model menu is hidden while Agent is working. ## What does Agent use credits for? Agent uses credits in two separate ways. | Use | When it is charged | Where you see it | | --------------- | ------------------------------------------------------------------------------ | ----------------------------------------------------------------------------- | | Agent reasoning | For each request, based on the Agent model and how much work the request takes | **Usage** tab of the [Tasks panel](https://docs.nodespell.com/graph/tasks/), listed as **Editor Agent** | | Generations | Only after you approve a run, like any run you start yourself | The Tasks panel, like other runs | Agent needs a small available credit balance before it can start a request. Each request also has a spending cap. If a request reaches the cap, Agent stops, and the usage entry shows **Budget reached**. Per-model prices change, so this page doesn't list them. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) for balances and plans. ## Stop, steer, queue, or resume a request While Agent is working, the message box changes. Its placeholder reads **Add a follow-up, or steer the current work**. | Control | What it does | | --------------------- | -------------------------------------------------------------------------------------------------------------------- | | **Stop** | Stops the current request. Completed changes are kept, and work already submitted may still finish. | | **Steer** | Appears when you have typed something. It stops the current work and continues with your new message instead. | | **Queue** | Sends your message after the current work finishes. Pressing `Enter` while Agent is working also queues the message. | | **Queued next** | Lists queued follow-ups in order. Remove one with its **×** button. You can queue up to eight. | | **Resume** | Appears after a stopped request. Agent rechecks the current graph and finishes only the unfinished steps. | | **Reconnect / Retry** | Appears if the connection was interrupted. It recovers Agent's saved answer. | 1. To redirect Agent mid-request, type the correction, for example "Use 9:16 instead". 2. Select **Steer**. Agent stops the current work and continues with your correction. 3. If you stopped too early, select **Resume** to finish the remaining steps. If Agent fails three times in a row, it pauses with **The assistant paused after repeated failures. Edit the prompt before trying again.** Rewrite the request before you send it again. Only one Agent request can run in a Project at a time. If you see **Another assistant request is already running in this project.**, stop that request or wait for it to finish. ## Frequently asked questions ### Does deleting a conversation undo Agent's changes? No. Deleting a conversation removes only its history. Graph edits and takes stay in the Project, and you undo edits in the graph. ### Which Agent model should I use? Start with the **Recommended** model in **Regular** for most graph work. Use **Cost effective** models for quick questions, and **High quality** models for long, complex tasks. ### Does Stop cancel a generation that already started? Not always. Work already submitted may still finish. Ask Agent to cancel the run, or cancel it in the Tasks panel. ### How many follow-ups can I queue? You can queue up to eight follow-ups while Agent is working. Agent sends them in order after the current request finishes. ## Related - [Agent overview](https://docs.nodespell.com/agent/) — open Agent and send a first request. - [Generate with Agent](https://docs.nodespell.com/agent/generate-with-agent/) — the credits a generation uses. - [Tasks panel](https://docs.nodespell.com/graph/tasks/) — the **Usage** tab and running work. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) — balances, plans, and estimates. - [What Agent can do](https://docs.nodespell.com/agent/what-agent-can-do/) — what a request can include. --- # What Agent can do in Nodespell > A complete reference of Nodespell Agent's capabilities, with an example request for each, the limits Agent works within, and tips for writing requests. Source: https://docs.nodespell.com/agent/what-agent-can-do/ · Last updated: 2026-11-10 Nodespell Agent works through a fixed set of actions. It can read your graph, search the catalogs, edit nodes and connections, work with Creative Tools and Assets, and run parts of the graph with your approval. This page lists every action group, with an example request for each, and the things Agent does not do. ## Agent capabilities at a glance In the table, "Changes the graph" means the action works only in **Edit** mode. "Needs approval" means Agent always asks first, even in **Edit**. See [Modes and approvals](https://docs.nodespell.com/agent/behavior-modes-and-approvals/). | What you can ask for | Example request | Changes the graph | Needs approval | | ------------------------------------------------------------------- | --------------------------------------------------------------- | ----------------- | -------------- | | Explain a graph, a node, or a chain | "Explain what this graph does." | No | No | | Find nodes on the current Level or anywhere in the Project | "Find every node that uses the portrait prompt." | No | No | | Trace connections upstream or downstream | "What feeds into @Final grade?" | No | No | | Count and summarize the graph | "How many image models are in this Project?" | No | No | | Find a node type or Creative Tool to use | "Find a faster video model for this branch." | No | No | | Read a node's or Creative Tool's settings and models | "What parameters does this upscaler have?" | No | No | | Create nodes | "Create an image node from this prompt." | Yes | No | | Duplicate a selected chain, up to 20 copies | "Duplicate these nodes three times with a different direction." | Yes | No | | Rename nodes | "Rename the shots Shot 01 to Shot 04." | Yes | No | | Change parameter values | "Set the aspect ratio to 9:16 on both image nodes." | Yes | No | | Connect or disconnect nodes | "Connect this audio into the video model." | Yes | No | | Delete nodes | "Delete the unused upscaler." | Yes | Yes | | Create a Creative Tool | "Add an image-to-video Creative Tool after @Hero still." | Yes | No | | Check what a model switch would drop | "What would I lose by switching this tool to another model?" | No | No | | Switch a Creative Tool's model | "Switch this tool to a cheaper model." | Yes | Yes | | Find Project Assets and place them on the graph, up to 20 at a time | "Add the three logo files from Assets to the graph." | Yes | No | | Check a node's results and review result media | "Did the last run of @Hero shot produce anything usable?" | No | No | | List recent runs and their status | "What's still running?" | No | No | | Estimate the credit cost of running nodes | "Find the most expensive step in this graph." | No | No | | Prepare a run with one to four outputs | "Get two takes of @Hero shot ready to run." | No | Yes | | Start a prepared run | "Start it." | No | Yes | | Follow a run's progress | "Tell me when the video finishes." | No | No | | Cancel a run | "Cancel the run you just started." | No | Yes | | Ask you a question with choices | Agent asks when your request is ambiguous. | No | No | Starting a run does not edit the graph, but it spends credits. For the full generation flow, see [Generate with Agent](https://docs.nodespell.com/agent/generate-with-agent/). ## What does Agent not do? These limits come from the actions Agent has. Agent can't work around them, even if you ask. | Agent does not | What to do instead | | -------------------------------------------------------- | -------------------------------------------------------------------------------------------- | | Spend generation credits without your approval | Approve the run when you're ready. | | Change the graph in **Ask** or **Plan** | Switch to **Edit**. | | Arrange or tidy the layout for you | See [Organize the graph](https://docs.nodespell.com/graph/organize/). Agent places new nodes automatically. | | Edit clips in the Timeline or layers inside a compositor | See [Timeline](https://docs.nodespell.com/edit/timeline/) and [Compositing](https://docs.nodespell.com/edit/compositing/). | | Use Snippets or Custom Nodes | See [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/). | | Import images or video from the web | Upload the file and attach it. See [Context and references](https://docs.nodespell.com/agent/context-and-references/). | | Retry a failed run by itself | Ask Agent to prepare the run again, and approve it. | | Listen to audio | Agent can read a transcript when one is available. | | Keep working if you switch Level or Project mid-request | Start a new request in the Level you want. | | Run two requests at once in one Project | Queue the follow-up, or wait. | Agent works only inside the open Project. It can't read or change other Projects. ## Tips for requests that work - **Name the target.** Include selected nodes, or mention a node with `@`, so Agent doesn't have to guess. - **Pick the mode first.** Use **Ask** for questions, **Plan** to review a change before it happens, and **Edit** to make it. - **Say the outcome and the constraint.** "Make a 9:16 version of @Hero shot and keep the prompt" works better than "make it vertical". - **Ask for an estimate before a big run.** "What will running this cost?" works in every mode and spends no generation credits. - **Give one task per message.** Queue the next step rather than putting several unrelated tasks in one request. - **Use attachment roles.** Mark a face reference as **Character** and a look reference as **Style**. - **Check the steps.** Each action Agent takes appears in the conversation. Select a node chip to find that node on the canvas. ## Frequently asked questions ### Can Agent build a whole workflow from a description? Yes, within the actions on this page. In **Edit** mode, Agent can create nodes and Creative Tools, set their parameters, and connect them. Running the result still needs your approval. ### Can Agent use any model in Nodespell? Agent searches the same node and Creative Tool catalogs you see in Nodespell. It uses models that are available to you. Running them still depends on your plan and credits. ### Can Agent see the images and videos it generated? Agent can check a node's results and inspect up to four results at a time, sampling a few frames from each video. It does not listen to audio. Treat Agent's comments on how a take looks as a starting point, and review the take yourself in the **Results** tab. ### Can Agent delete my nodes? Only with your approval. Agent asks before every deletion, and you can undo a deletion with `Cmd/Ctrl+Z`. ## Related - [Agent overview](https://docs.nodespell.com/agent/) — open Agent and send a first request. - [Modes and approvals](https://docs.nodespell.com/agent/behavior-modes-and-approvals/) — when Agent asks first. - [Context and references](https://docs.nodespell.com/agent/context-and-references/) — mentions and attachments. - [Generate with Agent](https://docs.nodespell.com/agent/generate-with-agent/) — estimates, runs, and takes. - [Creative Tools](https://docs.nodespell.com/creative-tools/) — the tools Agent can create and switch. --- # Find and inspect files in the Asset Manager > Use the Nodespell Asset Manager to browse uploaded and generated Assets, search and filter them, see details, and find the node that made each one. Source: https://docs.nodespell.com/assets/asset-manager/ · Last updated: 2026-11-10 The Asset Manager is the Nodespell panel for every file you keep: media you upload and results your nodes generate. In Nodespell these files are called Assets. You use the Asset Manager to browse, search, filter, tag and inspect Assets, drag them into a graph, and jump from a generated Asset back to the node that made it. ## Open the Asset Manager Open the account menu and choose **Assets**, or add the **Assets** panel from **Add New Panel** (in the **Essential** group). In Immersive mode, `T` shows or hides the Asset Manager. ## What is in the Asset Manager? | Area | What it holds | | ------------------ | ---------------------------------------------------------------------------------- | | Workspace selector | **Personal**, or one of your Teams. Each workspace has its own Assets. | | Quick views | **All Assets**, **Uploads**, **Generated**, **Current Project**, **Trash** | | **Browse by** | Switch the sidebar between **Folders** and **Tags** | | Toolbar | Search and filters, sort, **Grid view** / **List view**, **Inspector**, **Upload** | | Path bar | Where you are, with **Go Up** | | Asset grid or list | The Assets in the current view | | **Inspector** | Details of the selected Asset | When you open the Asset Manager from a Project, it selects that Project's workspace when it can. See [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/). ## Use the quick views | Quick view | Shows | | ------------------- | ---------------------------------------------------------------------------------------------------- | | **All Assets** | Everything in the workspace except Trash | | **Uploads** | Files you uploaded | | **Generated** | Results produced by nodes. While it is open, **Images**, **Videos** and **Audio** narrow it by type. | | **Current Project** | Assets associated with the open Project (available only with a Project open) | | **Trash** | Assets moved to Trash, with their recovery date | ## Search, filter and sort Type in **Search assets...** to match Asset names and folder paths. Search is not case-sensitive and does not search tags; use tag filters for that. Select the filter button at the end of the search field to open **Filters**: | Section | Filters | | -------- | ------------------------------------------------------------------- | | **Show** | **Recent · 7 days** (modified in the last seven days), **Untagged** | | **Type** | One chip per file type present, with a count | | **Tags** | One chip per tag present, with a count | The filter button shows how many filters are active. **Clear all** removes them. Sort by **Name**, **Modified** or **File Type**, ascending or descending. ## Save a view you use often At the bottom of **Filters**, under **Saved Views**, type a name and select **Save** to keep the current folder, search, sort and filters. New views are private. From a view's menu you can **Apply view**, **Pin** or **Unpin**, **Share with Team** or **Make private** (in a Team workspace), and **Delete**. ## Select Assets | Action | Result | | ----------------- | ----------------------------------------- | | Click | Select one Asset | | `Cmd/Ctrl`+click | Add or remove an Asset from the selection | | `Shift`+click | Select a range | | `Space` | Preview the selected Asset | | Double-click | Open the Asset in the full-screen viewer | | `Cmd/Ctrl+C` | Copy the URLs of the selected Assets | | `Esc` | Clear the selection | | Drag to the graph | Create Asset nodes for the dragged Assets | Right-click an Asset, or a selection, for its actions: **Open In New Tab**, **Info**, **Find in graph**, **Download Asset**, **Copy URL**, **Move to folder…**, **Manage Tags**, **Rename**, **Replace Asset**, and **Move to Trash**. ## What does the Inspector show? Select **Inspector** in the toolbar, or choose **Info**, to see the selected Asset: - Ownership: **Owner**, **Uploaded by**, **Storage charged to**, and the **Project** it belongs to, with **Reveal in Project**. - **Folder** and **Tags**, which you can edit in place with **Add tag...**. - **Media Info**: file size, MIME type, format, dimensions, duration, frame rate, bitrate, color space and alpha channel, where they apply. - **Technical Details**: name, path, modified time and URL, with **Copy URL**. ## Find the node that generated an Asset Generated Assets remember where they came from. Right-click a generated Asset and choose **Find in graph** to select the source node. If the Asset came from another Project, Nodespell asks **Open another Project?** before switching. If the node was deleted, you see **Source node not found**. **Restore generation workflow** rebuilds the nodes and connections that produced the Asset in a new group in the source Project. Existing nodes stay unchanged; review the restored inputs before running. ## Frequently asked questions ### Are generated results saved to the Asset Manager automatically? Yes. Results generated by nodes appear under **Generated** in the Project's workspace. Uploaded files appear under **Uploads**. ### Why is Current Project greyed out? **Current Project** needs an open Project. Open one and the view becomes available. ### Can search find an Asset by its tag? No. Search matches names and folder paths. Switch **Browse by** to **Tags**, or use the **Tags** filter chips. ### Who can replace or trash an Asset? In Personal, you can. In a Team, the uploader or the Team owner can; other members see **Only the uploader or Team owner can use this action**. ## Related - [Upload and organize Assets](https://docs.nodespell.com/assets/upload-and-organize/): add files, folders, tags and renaming - [Preview, download and Trash](https://docs.nodespell.com/assets/preview-download-and-trash/): preview, download, delete and restore - [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/): Personal vs Team Assets and storage limits - [Add nodes](https://docs.nodespell.com/graph/add-nodes/): drag Assets into a graph - [Supported media](https://docs.nodespell.com/reference/supported-media/): which files you can upload --- # Upload and organize Assets > Upload files to Nodespell from the Asset Manager, the graph or a node's asset picker, then organize Assets with folders, tags, renaming and moving. Source: https://docs.nodespell.com/assets/upload-and-organize/ · Last updated: 2026-11-10 Uploading in Nodespell turns a file from your computer into an Asset: a stored file you can reuse in any Project in the same workspace. You can upload from the Asset Manager, by dropping files onto the graph, or from a node's asset picker. Each file can be up to 200 MB, and you can upload up to 50 files at a time. ## Upload limits at a glance | Limit | Value | | ----------------------- | ---------------------------------------------------------------------------------------------------------- | | Size per file | 200 MB (larger files are refused with **Files too big, limit is 200MB**) | | Files per upload | 50 (more are refused with **Too many files, limit is 50**) | | Files uploading at once | 3; the rest wait as **Queued** | | Storage | Your workspace's allowance; see [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/) | | Accepted types | Images, video, audio, documents, archives and 3D files; see [Supported media](https://docs.nodespell.com/reference/supported-media/) | ## Upload from the Asset Manager 1. Open the [Asset Manager](https://docs.nodespell.com/assets/asset-manager/) and choose the workspace and folder you want. 2. Select **Upload** and choose files. 3. In the upload dialog, check the destination. It shows the workspace, the folder and **Storage charged to**. Pick another folder under **Upload location**, or select **Create New Folder**. 4. Select **Upload Assets**. Each file shows its status: **Queued**, **Uploading**, **Uploaded**, **Reused** or **Failed**. A failed file shows the reason; select **Retry** on it, or **Retry failed** for all of them. **Remove file** takes a file out of the queue before it uploads. The default destination is the **Uploads** folder. A folder you create in the upload dialog is temporary until a file is uploaded into it. Its name can use letters, numbers, spaces, hyphens and underscores. ## Upload by dropping files on the graph Drag files from your computer onto the Graph canvas. The upload dialog opens so you can choose the folder. When the upload finishes, Nodespell creates an Asset node for each file where you dropped them. Files over the limits, or of a type Nodespell does not recognize, are refused with **Unsupported file types** and the file names. You can also press `O` with the pointer over the canvas to open **Add asset**, then choose **Assets** (pick from your library), **Upload** (choose files), or **Load from URL** (paste a direct link to create an Asset node). Dropping a link creates an Asset node that points to that link; the file is not uploaded to your Assets. ## Upload from a node's asset picker When a node parameter asks for media, select **Choose assets**. The picker lets you choose existing Assets, upload files with **Choose files**, or add a public external URL. Uploaded files become Assets in the Project's workspace. ## What happens when I upload the same file twice? Nodespell recognizes identical file content in the same workspace. Instead of storing a second copy, the upload reports **Reused**, and the Asset Manager reveals the existing Asset in its current folder. The message reads, for example, **Reused 1 existing Asset**. A reused file does not use extra storage. Identical files in a different workspace (Personal and a Team) are separate Assets. ## Organize with folders | Action | How | | --------------- | ---------------------------------------------------------------------------------------- | | Create a folder | **New folder** for the top level, or **Create subfolder** from a folder's actions | | Rename a folder | **Rename folder**. Assets keep their files; only the Asset Manager organization changes. | | Move a folder | **Move folder**, then choose a destination or **Top level** | | Delete a folder | **Delete folder**. Only folders with no Assets and no child folders can be deleted. | | Move Assets | Select Assets, right-click, **Move to folder…**, choose the destination, **Move Assets** | Moving Assets only changes their folder inside the current workspace. It never changes who owns them or which storage they count against. To put an Asset in another workspace, copy it; see [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/). ## Organize with tags Tags let one Asset appear in several groupings. Nest tags with a slash, for example `client/acme`. - Select Assets, right-click and choose **Manage Tags**. Use **Add tag**, or type in **Search or create tags...** and choose **Create "…"**. - In the **Inspector**, type in **Add tag...** and select **Add**, or remove a tag from the list. - Switch **Browse by** to **Tags** to browse by tag. Each tag's options are **Rename**, **Color**, **Merge** and **Delete**. **Merge** moves every Asset tagged with one tag to another and previews how many Assets change. **Delete** removes the tag from every Asset; **Delete with children** also removes nested tags. Deleting a tag never deletes Assets. ## Rename or replace an Asset - **Rename**: right-click one Asset, choose **Rename**, and type the new name including the extension. - **Replace Asset**: right-click, choose **Replace Asset**, and pick a new file. In a Team, only the uploader or the Team owner can replace an Asset. ## Put Assets on the graph Drag Assets from the Asset Manager onto the Graph canvas to create Asset nodes, or onto a [Timeline](https://docs.nodespell.com/edit/timeline/)'s Media area. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/). ## Frequently asked questions ### Can I paste an image from the clipboard to upload it? Not into the Asset Manager or the graph. Use **Upload**, drag files onto the graph, or the asset picker. The [Agent](https://docs.nodespell.com/agent/context-and-references/) composer accepts pasted files as attachments. ### Why did my upload say Reused instead of Uploaded? The same file content already exists in this workspace. Nodespell points you to the existing Asset instead of storing a duplicate. ### Why was my upload refused for storage? The workspace has less free storage than the files need. See [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/) for how to free space. ### Can I upload a folder of files? Select the files inside the folder; you can choose up to 50 at a time. ## Related - [Asset Manager](https://docs.nodespell.com/assets/asset-manager/): browse, search and inspect Assets - [Preview, download and Trash](https://docs.nodespell.com/assets/preview-download-and-trash/): what to do with Assets after upload - [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/): Personal and Team storage - [Supported media](https://docs.nodespell.com/reference/supported-media/): accepted file types - [Add nodes](https://docs.nodespell.com/graph/add-nodes/): use Assets in a graph --- # Preview, download and restore Assets from Trash > Preview Nodespell Assets, download original files, move Assets to Trash, and restore them within the 30-day recovery period in the Asset Manager. Source: https://docs.nodespell.com/assets/preview-download-and-trash/ · Last updated: 2026-11-10 The Nodespell Asset Manager lets you preview any Asset, download its original file, and remove Assets you no longer need. Removing an Asset moves it to Trash rather than deleting it. Ordinary Assets stay recoverable in Trash for 30 days, and you can restore them to their folder during that time. ## Preview an Asset | How | What opens | | -------------------------------------------- | ---------------------------------------------- | | Select Assets and press `Space` | The full-screen viewer for the selection | | Double-click the Asset | The full-screen viewer for its type | | Right-click → **Open In New Tab** | The file in a new browser tab | | Right-click → **Info**, or the **Inspector** | Details, ownership, tags and media information | The full-screen viewers have their own playback, zoom and compare tools; see [Viewers](https://docs.nodespell.com/edit/viewers/). Some files have no browser preview. EXR images show **EXR preview unavailable** in the Asset Manager; the Asset itself is fine and can still be used in nodes and downloaded. Video Assets show a poster frame. If a video's thumbnail is missing, right-click it and choose **Generate Video Thumbnail**. ## Download Assets 1. Select one or more Assets. 2. Right-click and choose **Download Asset**, or **Download N assets** for a selection. 3. Nodespell starts one browser download per file and confirms **Downloads started**. Downloads are the original uploaded or generated files at full resolution, not previews. If some files fail, **Download failed** says how many. For other places to download from, see [Export finished media](https://docs.nodespell.com/edit/export/). ## Move Assets to Trash 1. Select the Assets you want to remove. 2. Right-click and choose **Move to Trash** (or **Move N assets to Trash**). 3. Read the message: the Assets move to this workspace's Trash and remain recoverable under its retention policy. 4. Select **Delete** to confirm. In Personal, you can trash your own Assets. In a Team, only the uploader or the Team owner can; for others the action is unavailable with **Only the uploader or Team owner can use this action**. Trashing an Asset does not change Projects that already use it, and it does not free storage straight away. The space is released when Nodespell permanently removes the Asset after the recovery period. ## Restore an Asset from Trash 1. Open the **Trash** quick view. 2. Find the Asset. Each one shows **Recoverable until** and a date. 3. Select **Restore**. The Asset returns to the folder it was in. If that folder no longer exists, it returns to **Uploads**. Only the uploader or the Team owner can restore an Asset; for others **Restore** is disabled with **Only the uploader or Team owner can restore this Asset.** ## How long do Assets stay in Trash? | Situation | What happens | | ------------------------------------------------------ | ------------------------------------------------- | | Ordinary Asset in Trash | Recoverable for 30 days, then permanently removed | | Asset still referenced or protected by retention rules | May be kept longer than 30 days | | Recovery date not yet known | Shows **Recovery expiry is being calculated.** | Each workspace has its own Trash. Personal Trash and each Team's Trash are separate. > **Caution:** The Asset Manager has no button to empty Trash or permanently delete an Asset immediately. Restore anything you need before its **Recoverable until** date. ## Frequently asked questions ### Does moving an Asset to Trash free storage? Not immediately. Assets in Trash still count toward storage until they are permanently removed after the recovery period. ### Will trashing an Asset break my Projects? Trashing does not edit Projects. Nodes that reference an Asset keep the reference, and Assets with active references may be kept beyond 30 days. ### Can I restore an Asset someone else uploaded to my Team? Only if you are the Team owner. Otherwise ask the uploader or the Team owner. ### Can I download several Assets as one ZIP? No. Nodespell downloads each selected Asset as its own file. ## Related - [Asset Manager](https://docs.nodespell.com/assets/asset-manager/): quick views, search and the Inspector - [Upload and organize Assets](https://docs.nodespell.com/assets/upload-and-organize/): folders, tags and renaming - [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/): what counts toward storage - [Export finished media](https://docs.nodespell.com/edit/export/): download from nodes, viewers and the Timeline - [Viewers](https://docs.nodespell.com/edit/viewers/): full-screen preview tools --- # Personal and Team Assets, and storage limits > Understand Personal and Team Asset workspaces in Nodespell, copy Assets between them, check your storage allowance, and free space when you hit the limit. Source: https://docs.nodespell.com/assets/workspaces-and-storage/ · Last updated: 2026-11-10 An Asset workspace in Nodespell is a separate Asset library with its own folders, tags, saved views, Trash and storage. Your **Personal** workspace is private to you. Each Team you belong to has one shared **Team** workspace for its members. Every Asset belongs to exactly one workspace, and its file counts toward that workspace's storage allowance. ## Personal vs Team Assets | | Personal | Team | | ------------------------------------------ | -------------------------- | ----------------------------------------------- | | Who sees the Assets | Only you | Members of that Team | | Folders, tags, Trash | Your own | Shared by the Team | | Saved views | Private | Private by default; can be shared with the Team | | Who can replace, trash or restore an Asset | You | The uploader or the Team owner | | Storage | Counts toward your storage | Counts toward the Team's storage | The Asset details show **Owner**, **Uploaded by** and **Storage charged to**, so you can see which workspace pays for each file. ## Switch Asset workspace Use the workspace selector at the top of the Asset Manager sidebar. It lists **Personal** and, under **Teams**, each Team you can access. When you open the Asset Manager from a Project, it selects that Project's workspace when it can. If you lose access to a Team while the panel is open, the Asset Manager returns to Personal and says **Team access changed. Asset Manager returned to Personal.** ## Where do uploads and generated results go? - Uploads go to the workspace and folder shown in the upload dialog. The dialog shows the workspace and **Storage charged to** before you upload. - Generated results go to the workspace of the Project that made them. A Team Project's results are Team Assets. ## Copy Assets between Personal and a Team **Move to folder…** only reorganizes Assets inside one workspace and never changes ownership. To put an Asset in another workspace, copy it. 1. Select the Assets in the Asset Manager. 2. Select **Copy to Team** (from Personal) or **Save personal copy** (from a Team) in the toolbar. 3. Choose the **Destination** and select **Preview copy**. The preview counts **Ready**, **Duplicates**, **Cannot copy** and **Storage required**. 4. Select **Copy Assets**. The summary reports how many were created, reused and failed. Copies are new, separate Assets. Projects that already use the source Assets keep pointing to the originals. Files that already exist in the destination are reused rather than copied again. A copy needs free storage in the destination; otherwise you see **The destination does not have enough storage for this copy.** ## How much storage do I have? Each plan includes a storage allowance for its workspace: | Plan | Storage | | ------------- | ------- | | Starter | 10 GB | | Pro | 25 GB | | Pro Plus | 25 GB | | Studio (Team) | 50 GB | Your account can differ from this table, for example after a plan change or a custom arrangement. To see your actual usage and limit, open **Account Settings** → **Subscription** → **Plan usage**. The **Storage** row shows what you use and your allowance, with a status of **Healthy**, **Approaching limit** or **Limit reached**. For plan details, see [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## What counts toward storage? - Every uploaded and generated Asset in the workspace, at its stored file size. - Assets in Trash, until they are permanently removed after the 30-day recovery period. - Re-uploading a file that already exists in the workspace does not add storage; the upload is **Reused**. ## What does "Storage limit reached" mean? The workspace does not have enough free space for the files you are adding. Nodespell checks before an upload, a copy, or a generation that saves new files. | Where you see it | Message | | ------------------------------- | ------------------------------------------------------------------------------------------------------------------------ | | Upload dialog, before uploading | **Selected files may exceed available storage (… left). Existing duplicate assets may still be reused.** | | Upload refused | **Storage limit reached. Delete assets or contact Nodespell to add more storage.** | | Run blocked | **Storage limit reached**: **Your workspace does not have enough storage for this action.** with **You're using … of …** | Existing Assets and Projects stay available. Only new files are blocked until you have room. ## Free up storage 1. Find large files: sort by **File Type** or filter by type, and check **File Size** in the Inspector. 2. Move Assets you do not need to Trash. Space is released when they are permanently removed after their **Recoverable until** date, not at the moment you trash them. 3. If you need space sooner or permanently, change plan, or contact Nodespell from the upload dialog's storage notice to add storage. ## Frequently asked questions ### Can I move an Asset from Personal to a Team without copying? No. Workspaces keep ownership separate. Copy the Asset with **Copy to Team**, then trash the Personal original if you no longer need it. ### Do Team Assets use my personal storage? No. Team Assets count toward the Team's storage. Check **Storage charged to** in the Asset details. ### Why didn't trashing Assets free space right away? Trashed Assets stay recoverable for 30 days and still count toward storage until Nodespell permanently removes them. ### Can I copy Assets from one Team to another Team? Not directly. From a Team workspace, the copy destination is your Personal workspace. ## Related - [Asset Manager](https://docs.nodespell.com/assets/asset-manager/): the workspace selector and Inspector - [Preview, download and Trash](https://docs.nodespell.com/assets/preview-download-and-trash/): Trash and restore - [Upload and organize Assets](https://docs.nodespell.com/assets/upload-and-organize/): uploads and duplicates - [Teams](https://docs.nodespell.com/teams/): what a Team is and switching workspace - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): plan allowances --- # Find, create and manage projects on the dashboard > Use the Nodespell projects dashboard to search, sort, create, open, rename, duplicate, archive and delete Projects in personal and Team workspaces. Source: https://docs.nodespell.com/projects/dashboard/ · Last updated: 2026-11-10 The projects dashboard is the Nodespell home screen where you find, create and manage your Projects. A Project is a saved, editable workspace that holds a graph and its panel layout. The dashboard lists the Projects in your current workspace, personal or Team, and gives each one a card with actions for opening, renaming, duplicating, archiving and deleting it. ## What is on the dashboard? The left sidebar of the dashboard controls what the main area shows. | Sidebar item | What it shows | | ----------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------ | | Workspace switcher (top) | Your current workspace. Choose **Personal workspace** or one of your **Team Workspaces**, or **Create A Team**. | | **Projects** | Active Projects in your personal workspace. | | **Favourites** | Personal Projects you starred. | | **Shared projects** / **My drafts** | In a Team workspace: Projects shared with the Team, and drafts only you can see. | | **Assets** | Your Assets. See [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). | | **Publishing** | **Workflow Submissions** and **Spellworks**. See [Templates and sharing](https://docs.nodespell.com/projects/templates-and-sharing/). | | **More** | **What's new**, **Language**, **Copy support email**, **Discord**, **YouTube channel**, **Archived projects** and **Visit website**. | | **Account** | **Messages**, **Subscription** (or **Team billing** in a Team workspace), **Refer and Earn**, and your account menu. | You can collapse the sidebar to icons, or drag its edge (**Resize sidebar**) to change its width. ## Find a project with search, tags and sorting The toolbar above the project list helps you narrow a long list. - **Search projects** filters by project name. Press `/` anywhere on the dashboard to jump to the search field. - **Tags** filters by one or more project tags. **Clear filters** removes every filter when nothing matches. - **Sort by** orders the list by **Last Modified**, **Date Created** or **Name**. Dates sort **Latest First** or **Oldest First**; names sort **A to Z** or **Z to A**. - **Layout** switches between a grid of cards and a list. The list layout adds tags and sortable column headers. The toolbar also shows how many projects are loaded. Long lists load more cards as you scroll. ## Create a new project 1. Choose the workspace you want the Project to belong to with the workspace switcher. 2. Select **New Project**. In a Team workspace the button says **New Shared Project** in **Shared projects** and **New Draft** in **My drafts**. 3. In the **New Project** dialog, enter a **Name** and, if you like, add **Tags**. 4. Select **Save**. Nodespell creates the Project and opens it. Inside an open Project you can also choose **File** > **New Project**. Two Projects in the same workspace cannot share a name; the dialog reports **Name clash** if you try. If your plan's active-project limit is full, the dialog shows **Active project limit reached**. Archive a Project, upgrade, or contact Nodespell to create more. Plan limits are listed in [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Open a project Select a card, or focus it and press `Enter` or `Space`. You can also right-click a card, or use its **More actions** button, and choose: - **Open** to open the Project for editing. - **Open Read-Only** to inspect the Project without making changes. Inside a Project, **File** > **Open ...** returns to the project browser, and **File** > **Recent Projects** lists Projects you opened recently in this browser. **Clear Recents** empties that list. A Project can be open in more than one browser tab, but only one tab edits it at a time. See [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/). ## Rename, retag or restyle a project Right-click a card, or use **More actions**, and choose: | Action | What it does | | ----------------------- | --------------------------------------------------------------------------------------------------------------------------- | | **Edit** | Opens **Edit Project**, where you change the **Name** and **Tags**. | | **Cover Colour** | Sets the card background to a preset colour, a **Custom Colour...** (single colour or gradient), or **Reset Cover Colour**. | | **Thumbnail** | **Upload Image...**, **Use Image URL...**, **Use Graph Preview**, or **Refresh Thumbnail**. | | **Copy URL To Project** | Copies a link to the Project. Only people with access to the Project can open it. | | **Add To Favourites** | Stars the Project so it appears in **Favourites**. | You can also rename the open Project from the project name in the top bar. ## Duplicate a project Choose **Duplicate** from a card's menu. Nodespell creates a copy in the same workspace with `_copy` added to the name. To copy the Project you have open under a name you choose, use **File** > **Save As**; this creates a new Project and does not rename the current one. ## Archive, restore and delete projects Archiving takes a Project out of your active list without deleting it, and frees a place in your active-project allowance. 1. Choose **Archive** from the Project's menu. 2. To find it again, open **More** > **Archived projects**. 3. Choose **Restore** to make it active again. If an active Project already uses the same name, the restored Project gets a new name that includes the restore time. **Delete** appears only on archived Projects. The confirmation reads **Are you absolutely sure?** and warns that the Project and its contents are permanently deleted. > **Caution:** Deleting a project cannot be undone from the app. Archive a Project instead if you might need it later. ## Move a personal project into a Team From a personal Project's menu, choose **Transfer Project**, pick the Team, then slide **Transfer To This Team** to finish. The Project becomes a Team Project shared with that Team. See [Team projects](https://docs.nodespell.com/teams/team-projects/). ## Frequently asked questions ### Can I undo deleting a project? No. **Delete** is only offered for archived Projects, and the confirmation states that it cannot be undone. Restore the Project from **Archived projects** instead if you may need it. ### Why can I not create a new project? Your workspace has reached its active-project limit, and the dialog shows **Active project limit reached**. Archive a Project, upgrade your plan, or contact Nodespell. ### Does archiving a project delete its media? Archiving only moves the Project out of the active list, and **Restore** brings it back. Asset retention follows your plan, as described in [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ### Where do my recent projects come from? **Recent Projects** in the **File** menu lists Projects recently opened in this browser. It is not shared between devices, and **Clear Recents** empties it. ## Related - [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/): save named versions of a Project. - [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/): how Nodespell saves your work. - [Teams overview](https://docs.nodespell.com/teams/): switch between personal and Team workspaces. - [Team projects](https://docs.nodespell.com/teams/team-projects/): shared Projects and drafts. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): active-project limits by plan. --- # Save, open and restore project versions > Learn how Nodespell Projects and Project Snapshots work, and how to save a named version, view it read-only, edit it, delete it or restore it. Source: https://docs.nodespell.com/projects/projects-and-snapshots/ · Last updated: 2026-11-10 A Project Snapshot is a named, saved version of a Nodespell Project that you can return to later. A Project is the editable workspace that holds your graph and panel layout; it saves automatically as you work. A snapshot captures the Project at one moment so you can view that state read-only, compare it with the live Project, or restore it. ## How are Projects and Project Snapshots different? | | Live Project | Project Snapshot | | ---------------- | ---------------------------------------- | -------------------------------------------------------------------------------------- | | What it is | The current, editable state of your work | A named copy of the Project at one moment | | How it is saved | Automatically, as you edit | Only when you choose **Save Version ...** | | Can you edit it? | Yes, in the tab that is editing | No, a snapshot opens read-only | | Where it appears | Projects dashboard | **File** > **Versions** inside the Project, and the snapshot count on the project card | Nodespell does not create snapshots for you. Automatic saving of the live Project is covered in [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/). ## What a snapshot keeps A snapshot records the Project's nodes, connections, parameters, layout and stored previews, plus which [take](https://docs.nodespell.com/graph/takes/) each node is using. It refers to existing generated media rather than making separate copies of it. If a take used by a snapshot is later deleted, that result can be unavailable when you open the snapshot. When you open a snapshot, node history shows only results that had finished by the time the snapshot was saved. ## Save a version of your project 1. Make sure you are on the live Project and editing it. If a change is still in progress, finish it first. 2. Open **File** and choose **Save Version ...**. 3. In **Save snapshot**, enter a **Name**, pick a **Color**, and optionally add a **Comment** (for example, what changed or why you saved it). 4. Select **Save**. A progress percentage shows while the version uploads, then **Snapshot saved** appears. If saving fails, the message says your project is unchanged. Common messages include **Finish your current edit, then save the version again.**, **Another version is being saved or restored. Try again in a moment.**, and **Versions can only be saved from the live project while you’re editing it.** ## Open and browse saved versions 1. Open **File** and point to **Versions**. Each snapshot is listed with its colour, name and how long ago it changed, newest first. 2. Select a snapshot. The Project switches to that version in read-only mode, and the version bar appears. 3. Inspect nodes, parameters and takes. You can copy settings, but you cannot change the snapshot. 4. Choose **File** > **Go To Live Version** to return to the live Project. Use **View** > **Show Version Bar** or **Hide Version Bar** to toggle the bar. If a version cannot open completely, Nodespell can offer **View available data**. This opens a partial, read-only view with a warning banner and a searchable list of the records it could verify, each with **Copy settings**. A partial view cannot be restored or published. ## Rename or recolour a snapshot While a snapshot is open, choose **File** > **Versions** > **Edit**. In **Edit snapshot**, change the **Name**, **Color** or **Comment**, then select **Update**. ## Restore a snapshot Restoring makes a snapshot the live version of the Project again. 1. Open the snapshot from **File** > **Versions**. 2. Choose **File** > **Versions** > **Restore ...**. 3. Read the confirmation, then select **Restore this version**. The confirmation explains that the live version is replaced and that changes made since the snapshot will be lost. Save a version of the current state first if you might want it back. In some cases Nodespell reviews the restore first, shows how many nodes are added, changed and removed, and saves a version named “Before restoring _name_” so you can return to it. Restoring does not delete any node's result history. New runs after the restore appear in history as normal. ## Delete a snapshot While a snapshot is open, choose **File** > **Versions** > **Delete ...**, then **Delete version**. The live Project is not affected. > **Caution:** Deleting a snapshot is permanent. Restoring a snapshot replaces the live Project. Check you have the right version selected before either action. ## How long are snapshots kept? Snapshot retention depends on your plan and is shown as **Snapshot retention** under **Plan usage** in your subscription settings. | Plan | Snapshot retention | | ------- | ------------------ | | Starter | 30 days | | Pro | 1 year | | Studio | 2 years | See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) for the full plan comparison. ## Frequently asked questions ### Does Nodespell save versions automatically? No. Your live Project saves automatically, but snapshots are created only when you choose **File** > **Save Version ...**. ### Can I edit a snapshot? No. A snapshot opens read-only. To work from it, restore it with **Restore ...**, or return to the live Project and make your changes there. ### Does deleting a snapshot change my project? No. **Delete version** removes only that saved version. The live Project is not affected. ### Why is Save Version unavailable? **Save Version ...** works only on the live Project while your tab is editing it. It is disabled while you are viewing a snapshot. ## Related - [Projects dashboard](https://docs.nodespell.com/projects/dashboard/): create, open and archive Projects. - [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/): how the live Project saves. - [Takes](https://docs.nodespell.com/graph/takes/): the results a snapshot points to. - [Team projects](https://docs.nodespell.com/teams/team-projects/): versions in shared Projects. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): retention by plan. --- # How saving works and how to recover unsaved edits > Understand how Nodespell saves Projects automatically, what the save status means, and how to review, save, export or discard edits that could not save. Source: https://docs.nodespell.com/projects/saving-and-recovery/ · Last updated: 2026-11-10 Nodespell saves every Project automatically as you edit; there is no Save button for the live Project. A status in the project bar tells you whether your latest changes are saved, still saving, or need your attention. If a change cannot be saved, Nodespell keeps it on your device so you can review it, save it again, export it, or discard it. ## What does the save status mean? The save status sits in the project bar at the top of the workspace. Quiet states are plain text; states that need attention are coloured pills. Hover or open a pill for details. | Status | Meaning | | ------------------- | ------------------------------------------------------------------------------------- | | **Saved** | Your latest changes are saved. | | **Saving…** | Your latest changes are being saved. | | **Changes pending** | You are still adjusting a value, such as dragging a slider. It saves when you finish. | | **Retrying** | Nodespell could not reach the server and is retrying automatically. | | **Not saved** | Some changes are not saved yet. Keep the tab open. | | **Needs action** | The server refused some changes. Open the pill to retry or export them. | While the connection is healthy, Nodespell also keeps a backup of unsaved changes on your device, and the details say **A backup is kept on this device.** The backup gets its own pill only when something is wrong: | Backup pill | Meaning | | ----------------------- | ---------------------------------------------------------------------------- | | **Saved locally** | Changes are kept on this device and waiting for the server. | | **Not backed up** | Changes are not protected locally. Keep the tab open until they save. | | **Uploads pending** | Media uploads are incomplete. The local backup does not include those files. | | **_n_ unsaved** | Unsaved actions are building up. Editing pauses at 99. | | **Editing paused** | 99 actions are waiting to save. Editing resumes when saves finish. | | **Needs attention** | Some kept edits need your review. Their save outcome may be unknown. | | **Save status unknown** | Nodespell could not check the kept edits. Keep the tab open. | ## What happens when you go offline? When your connection drops, Nodespell shows **You're offline** and changes cannot be saved until you are back online. Keep the tab open: edits stay in the tab and in the local backup, and saving resumes when the connection returns. Temporary interruptions recover without you making another edit. ## Review recovered edits When some edits could not be saved, they stay on this device until you decide what to do. Open **Needs attention**, or the **Needs action** pill, and choose **Review changes** to see the **Unsaved changes** list. Each change shows what it was (for example **Moved**, **Renamed**, **Connected** or **Picked a take**), which nodes it affects, and its status, such as **The server refused this**, **May already be saved** or **Not sent yet**. | Action | What it does | | ------------------------------------ | ----------------------------------------------------------------------- | | **Show** | Selects the affected node in the graph. | | **Use mine** / **Use all mine** | Puts your recorded version back into the Project. | | **Save mine** / **Save _n_ changes** | Saves the changes that can be saved automatically, together. | | **Discard** / **Discard all** | Removes the change from this device. The saved Project is not affected. | | **Export for support** | Downloads a recovery file you can send to Nodespell. | Some changes cannot be put back automatically and say **Redo by hand**. For these, the list shows **Your values, as recorded** with a copy button for each field, so you can redo the change yourself. ## How do you know a recovered edit is saved? After you save a recovered edit, Nodespell checks it against the Project. Confirmed changes move to **Saved**, and when everything is confirmed the list shows **All changes are saved**. If only part of a recovery succeeds, the saved parts move to **Saved** and the rest stay listed for review, for example **Save not confirmed** or **Project refused it**. **Check status** checks again. A refused change does not block other edits that can save safely. You can keep working while it waits for review. ## Export a copy of unsaved work Use **Export copy** in the save or recovery panel to download a recovery file of the edits that are not saved. Export before you discard anything you might need. If an export fails, keep the tab open and try again. ## Discard unsaved changes 1. Open the recovery panel and choose **Discard changes**. 2. Read **Discard retained edits?**. It explains that all retained local edits for this Project are removed and the saved Project reopens. 3. Select **Discard and reload**, or **Keep changes** to cancel. Changes that were already saved on the server stay. Discarding also removes edits that were waiting behind a refused one. > **Caution:** Discarded edits cannot be brought back. Use **Export copy** first if you may need them. ## What happens with two tabs or two devices? Only one browser tab edits a Project at a time. The project bar shows the editing state: | Status | Meaning | | --------------------- | ----------------------------------------------------------------------------------------------------------------- | | **Editing** | This tab is editing the Project. | | **Viewing** | Another tab of yours is editing. Click anywhere to edit here; the other tab saves its changes first, then pauses. | | **_name_ is editing** | A teammate is editing. Choose **Request editing** to ask them to hand over. | | **Paused** | This tab paused after a while without activity. Move the pointer or press a key to continue. | | **Reconnecting…** | Nodespell is reconnecting. Your changes are safe on this device. | While viewing, you can still browse Assets and inspect media. If your other tab or device stops responding, Nodespell offers **Take over now**. Anything that tab had not saved stays in it and does not appear in the new tab, so you can still export it there. ## Closing a project with unsaved changes When you close a Project, Nodespell first finishes saving. If some changes have not finished saving, the Project stays open with **Project not closed**, and nothing is lost. Check the recovery notice, then try again. ## Frequently asked questions ### Do I need to save my project? No. Nodespell saves the live Project automatically. Use **File** > **Save Version ...** only when you want a named [snapshot](https://docs.nodespell.com/projects/projects-and-snapshots/). ### Is it safe to close the tab while it says Not saved? No. **Not saved** means some changes have not reached the server. Keep the tab open until the status returns to **Saved**, or export a copy first. ### Why is editing paused? Editing pauses when 99 actions are waiting to save, so unsaved work cannot grow without limit. It resumes when saves finish. ### Does discarding changes affect saved work? No. Discarding removes only the unsaved edits kept on this device. Changes already saved on the server remain. ### Can two people edit the same project at once? Only one tab edits a Project at a time. Another person can request editing, and the current editor chooses whether to hand over. See [Team projects](https://docs.nodespell.com/teams/team-projects/). ## Related - [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/): save named versions. - [Projects dashboard](https://docs.nodespell.com/projects/dashboard/): open and manage Projects. - [Team projects](https://docs.nodespell.com/teams/team-projects/): editing a shared Project. - [Support](https://docs.nodespell.com/account/support/): what to send when something goes wrong. - [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/): fixes for common problems. --- # Share projects and submit workflows to Nodespell > Share Nodespell Projects through Teams and project links, reuse published Workflows, and submit Workflows or Spellworks videos for Nodespell review. Source: https://docs.nodespell.com/projects/templates-and-sharing/ · Last updated: 2026-11-10 Nodespell gives you three ways to share work: share a Project with a Team, reuse published Workflows as starting points, and submit your own Workflows or finished videos to Nodespell for review and publication. A Workflow is a saved graph, from two nodes to hundreds, that opens as visible, editable structure. Nothing becomes public unless you submit it and Nodespell publishes it. ## How do I share a project with someone? Share a Project by putting it in a [Team](https://docs.nodespell.com/teams/) that the other person belongs to. Every member of a Team can open and edit the Team's shared Projects, one editor at a time. | Where the Project lives | Who can open it | | ---------------------------- | ------------------------------------- | | Your personal workspace | Only you | | A Team's **Shared projects** | Every member of that Team | | A Team's **My drafts** | Only the person who created the draft | **Copy URL To Project** on a project card copies a direct link. The link opens the Project only for people who already have access to it; it does not grant access by itself. To share a personal Project, move it into a Team with **Transfer Project**. See [Team projects](https://docs.nodespell.com/teams/team-projects/). ## Start from a published Workflow Published Workflows are ready-made graphs you can open and edit. In the editor, open the **Library** panel and choose **Workflows**. The **Source** filter offers **All sources**, **Official**, **Mine** and **Favourites**. When you open a published Workflow from a link or a public Workflow page, Nodespell creates a new Project from it, copies the Workflow and its sample results in, and opens it. A confirmation such as “Workflow copied to _project name_” appears. If copying fails, the Workflow may have been unpublished or may contain Assets that are no longer available. To save part of your own graph as a reusable Workflow, see [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/). ## Submit a project or Workflow to Nodespell You can submit a whole Project, or a Workflow from your Library, for Nodespell to review and publish. 1. In an open Project, choose **File** > **Submit Project to Nodespell**. For a Workflow, open its detail page in the Library and choose **Submit to Nodespell**. 2. In **Submit to Nodespell**, fill in **Public title**, **Public description**, **Category** and **Tags**. 3. Under **Thumbnail**, choose a thumbnail Asset. Under **Example media**, add at least one example. These are picked from your Asset Manager and copied into the public publication if it is published. 4. Optionally add **Notes to reviewers**. 5. Tick the rights confirmation: you own or have the rights to all Assets, media, prompts and outputs in the Workflow. 6. Select **Submit**. Every field above except **Notes to reviewers** is required. **Submit Project to Nodespell** is unavailable while you are viewing a snapshot. A Workflow can only have one active submission at a time. ## Track your submissions After submitting, **Submission sent** offers **View submissions**. You can also open **Publishing** > **Workflow Submissions** in the dashboard sidebar at any time. | Status | Meaning | | ----------------------------- | -------------------------------------------------------------- | | **Draft submission** | Saved but not yet sent. | | **Submitted** / **In review** | Waiting for or under Nodespell review. | | **Approved** / **Published** | Accepted; **Published** Workflows have a public page. | | **Rejected** | Not accepted. Read **Reviewer notes**, then edit and resubmit. | | **Withdrawn** | You took it out of review. | Depending on the status, a submission offers **Public page**, **Use**, **Edit and resubmit**, **Withdraw** and **Delete**. **Delete** removes the submission and the public copies of its Assets; your original Workflow or Project is kept. ## Submit a video to Spellworks Spellworks is where Nodespell showcases finished work made with Nodespell. 1. In the dashboard sidebar, open **Publishing** > **Spellworks**. 2. Select **Submit video**. 3. Upload the final video file (recommended, so Nodespell can review and preserve it) or link to where it is hosted. 4. Add a description, tags, **Work links** for the piece itself, and any **Behind-the-scenes links** such as making-of posts. 5. Submit for review. The **Spellworks submissions** list shows each submission's status, with **Edit**, **Resubmit**, **Withdraw** and **Public page** when they apply. A linked video depends on its original host staying available. ## How your name appears on published work Published Spellworks, Workflows and Projects show your public creator profile. Set your public display name, creator URL, bio and links in **Account Settings** > **Creator Profile**. See [Account settings](https://docs.nodespell.com/account/settings/). ## Frequently asked questions ### Does Copy URL To Project make my project public? No. The link opens the Project only for people who already have access to it. Share a Project by moving it into a Team. ### Is my project public after I submit it? No. A submission goes to Nodespell review first. It becomes public only when its status is **Published**. ### What happens to my original project when I delete a submission? Your original Workflow or Project is kept. Deleting a submission removes the submission and the copies of Assets made for its public page. ### Can I edit a published Workflow I opened? Yes. Opening a published Workflow creates your own new Project with an editable copy. Changes you make do not affect the published Workflow. ## Related - [Team projects](https://docs.nodespell.com/teams/team-projects/): share Projects with a Team. - [Projects dashboard](https://docs.nodespell.com/projects/dashboard/): copy links and transfer Projects. - [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/): save reusable Workflows. - [Account settings](https://docs.nodespell.com/account/settings/): your public creator profile. - [Asset Manager](https://docs.nodespell.com/assets/asset-manager/): choose thumbnails and example media. --- # Teams in Nodespell > Learn what a Nodespell Team is, how to create one, which plans let you add members, and how to switch between your personal and Team workspaces. Source: https://docs.nodespell.com/teams/ · Last updated: 2026-11-10 A Team in Nodespell is a shared workspace where invited members work on the same Projects. Each Team has one owner and any number of members up to its plan's limit. A Team has its own shared Projects, private drafts and Team Assets, kept separate from your personal workspace, and its credits come either from each member or from one shared pool. ## Personal workspace and Team workspaces Every account has a personal workspace. When you belong to a Team, you also have that Team's workspace. | | Personal workspace | Team workspace | | ---------------------- | -------------------------------------- | ---------------------------------------------------------------- | | Description in the app | “Your own projects, credits, and plan” | “Shared team projects and team pool credits” | | Projects | Only yours | **Shared projects** for every member, plus **My drafts** for you | | Assets | Personal Assets | Team Assets | | Credits used | Your personal credits | Your personal credits, or the Team's shared pool on Studio | Personal and Team credits are never combined. See [Team credits and usage](https://docs.nodespell.com/teams/credits-and-usage/). ## Switch between workspaces 1. On the projects dashboard, open the workspace switcher at the top of the sidebar. It shows your current workspace name. 2. Under **Current workspace** you see the workspace's description and which credits it uses, such as “Uses Pro credits” or **Studio pool**. 3. Choose **Personal workspace**, or a Team under **Team Workspaces**. The project list, the **New Project** button and the sidebar items change to match the workspace. In a Team workspace the sidebar shows **Shared projects**, **My drafts** and **Team actions**. ## Which plans include Teams? | Plan | What you can do with Teams | | ------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Free, Starter | Create one Team, but you cannot add members until you upgrade. | | Pro, Pro Plus | Teams with up to 3 team members. Each member uses their own credits. | | Studio | A Team with one shared credit pool paid centrally by the owner. Studio starts at 2 team members; self-serve checkout goes up to 4, and larger Teams are set up with Nodespell, up to 10. | Team member places count everyone who has joined, including the owner, plus pending invitations. Plan details are in [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Create a Team 1. Open the workspace switcher and choose **Create A Team**. You can also open **Account Settings** > **Teams** and select **Create team**. 2. Enter a **Team Name**. Optionally set an **Icon** (initials or an emoji), a **Color**, or **Upload Avatar**. 3. Select **Create team**. Nodespell then opens the **Teams** tab with a setup checklist for the new Team: **Choose a plan**, **Invite your team** and, on Studio, **Review pool alerts**. The setup asks **How will credits work in** your Team, with two choices: - **Everyone uses their own credits**: the Pro plan, up to 3 team members. - **One shared pool, paid centrally**: the Studio plan. ## Manage a Team Open **Account Settings** > **Teams**, or choose **Members** under **Team actions** in a Team workspace. Pick the Team from **Select a team**. The page has these tabs: | Tab | Who sees it | What it contains | | ---------------- | ------------------------------ | ------------------------------------------------- | | **Members** | Everyone | Members, invitations and team capacity | | **Credit usage** | Owner; members of Studio Teams | Team credit use and the shared balance | | **Billing** | Owner | How the Team pays and **Manage plan and top-ups** | The owner's **Team settings** button offers **Edit name and appearance**, **Delete team** and **Hand over this team**. Members see **Leave team** instead. ## Frequently asked questions ### Can I create a Team on the Free plan? Yes. Any plan can create one Team, but adding members requires Pro or Studio. The app shows **Add team members with Pro or Studio** until you upgrade. ### Do my personal credits pay for Team work? On a Pro Team, yes: each member's own credits pay for their runs in the Team. On a Studio Team, work in the Team workspace uses the Team's shared pool, and your personal credits stay separate. ### Can a Team have more than one owner? No. Each Team has exactly one owner. To hand a Team to someone else, contact support with the new owner's email, because billing moves too. ### Can I belong to more than one Team? Yes. Every Team you belong to is listed under **Team Workspaces** in the workspace switcher. ## Related - [Members and roles](https://docs.nodespell.com/teams/members-and-roles/): Invite people and see who can do what. - [Team projects](https://docs.nodespell.com/teams/team-projects/): Shared projects, drafts and editing. - [Team credits and usage](https://docs.nodespell.com/teams/credits-and-usage/): Shared pools, usage and alerts. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): Which plan includes which Team features. --- # Invite team members and understand roles > Invite people to a Nodespell Team, accept or decline invitations, remove members or leave a Team, and see exactly what owners and members can do. Source: https://docs.nodespell.com/teams/members-and-roles/ · Last updated: 2026-11-10 A Nodespell Team has two roles: one owner and any number of members up to the plan's limit. The owner invites and removes people, manages billing and can delete the Team. Members work on the Team's shared Projects. Invitations go to an email address, last seven days, and must be accepted by someone signed in with that verified email. ## What can owners and members do? | Action | Owner | Member | | ----------------------------------------------------------------------------- | ------------------------------ | ------------------------------ | | Open and edit **Shared projects** | Yes | Yes | | Create shared Projects and private drafts | Yes | Yes | | See another person's drafts | No | No | | Switch own Team Project between **Share With Team** and **Move To My Drafts** | Yes, for Projects they created | Yes, for Projects they created | | Archive or delete a shared Project | Yes | Yes | | Transfer one of their personal Projects into the Team | Yes | Yes | | See the member list | Yes | Yes | | Invite, resend or cancel invitations | Yes | No | | Remove a member | Yes | No (members can leave) | | See Team credit use by day and member, download CSV | Yes | No (members see their own use) | | Turn **Pool alerts** on or off | Yes | No | | Top up the Team's shared pool and manage the Team plan | Yes | No (members can ask the owner) | | Edit the Team name and appearance | Yes | No | | Delete the Team | Yes | No | | Leave the Team | No | Yes | Roles cannot be changed in the app. To make someone else the owner, use **Hand over this team** in **Team settings** and contact support, because billing moves with ownership. ## Invite someone to your Team You must be the Team owner, and the Team's plan must include team members. 1. Open **Account Settings** > **Teams**, or choose **Members** under **Team actions** in the Team workspace. 2. Select the Team in **Select a team**. 3. On the **Members** tab, under **Invite a team member**, enter the person's email address. 4. Select **Send invitation**. The person does not need an account yet; they can create one after receiving the invitation. Invitations are listed with **Pending** or **Expired** after the email address, and the capacity line shows how many of the Team's places are used, counting pending invitations. | Message | What it means | | -------------------------------------------------------- | ------------------------------------------------------- | | **This person is already a team member.** | They have already joined. | | **An invitation has already been sent to this address.** | A pending invitation exists. Use **Resend** instead. | | **Team member limit reached.** | All places are used by members and pending invitations. | | **Upgrade to enable team members.** | The Team's plan does not include members. | ## Resend or cancel an invitation Invitations expire after seven days. Select **Resend** on a pending or expired invitation to send it again; this gives it another seven days. You must wait one minute after sending before you can resend. Cancel an invitation from the same list to free its place. ## Accept or decline an invitation When you are invited, Nodespell shows **You've been invited to a team** with a **Review** button, and the invitation is listed under **Invitations** in **Account Settings** > **Teams**. 1. Sign in with the email address the invitation was sent to, and make sure it is verified. 2. Choose **Review invitation**. 3. Read who invited you and how credits work in that Team. 4. Select **Join team**, or **Decline**. If you see **This invitation is unavailable for your account**, sign in with the invited address and verify it. An expired invitation must be resent by the Team owner. ## Remove a member As the owner, open the **Members** tab and remove the person, then confirm. They lose access through the Team; the Team's Projects and Assets stay. On Studio, their place becomes free, and your bill and monthly credits do not change unless you change your plan. ## Leave a Team As a member, open the Team in **Account Settings** > **Teams**, select **Leave team**, type `leave` to confirm, then select **Leave Team**. You need a new invitation from the owner to rejoin. The owner cannot leave their own Team. ## Frequently asked questions ### Does the owner count as a team member? Yes. The member limit counts everyone who has joined, including the owner, plus pending invitations. ### Can a member invite other people? No. Only the Team owner can send, resend or cancel invitations. ### Can members see each other's drafts? No. Drafts in **My drafts** are visible only to the person who created them, including to the Team owner. ### What happens to my work when I am removed from a Team? You lose access to the Team's Projects and Assets. The Team's Projects and Assets stay with the Team. ### How long does an invitation last? Seven days. The owner can resend it, which gives it another seven days. ## Related - [Teams overview](https://docs.nodespell.com/teams/): create a Team and switch workspaces. - [Team projects](https://docs.nodespell.com/teams/team-projects/): shared Projects, drafts and editing. - [Team credits and usage](https://docs.nodespell.com/teams/credits-and-usage/): who pays for runs. - [Sign-in and security](https://docs.nodespell.com/account/sign-in-and-security/): verifying your email. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): member limits by plan. --- # Work on shared Team Projects > Create shared Team Projects and private drafts in Nodespell, move personal Projects into a Team, and understand how one-editor-at-a-time editing works. Source: https://docs.nodespell.com/teams/team-projects/ · Last updated: 2026-11-10 A Team Project is a Nodespell Project that belongs to a Team instead of to one person. Every member can open and edit the Team's shared Projects, while drafts stay private to the person who created them. Only one person, in one browser tab, edits a Project at a time; others view it and can ask to take over. ## Shared projects and drafts In a Team workspace, the dashboard sidebar shows two lists. | List | Who sees it | Button to create one | | ------------------- | ------------------------ | ---------------------- | | **Shared projects** | Every member of the Team | **New Shared Project** | | **My drafts** | Only you | **New Draft** | When you create a draft, the dialog notes that the new draft stays private to you inside the Team. Neither the Team owner nor other members can open your drafts. ## Share a draft with the Team, or make a Project private The person who created a Team Project can change who sees it. Right-click the project card, or open **More actions**, and choose: - **Share With Team** to move a draft into **Shared projects**. - **Move To My Drafts** to make a shared Project private to you again. Only the Project's creator sees these actions. ## Move a personal Project into a Team 1. Switch to your personal workspace and find the Project. 2. Right-click the card, or open **More actions**, and choose **Transfer Project**. 3. Pick the Team. 4. Slide **Transfer To This Team** to finish. The Project leaves your personal workspace and appears in the Team's **Shared projects**. You must be a member of the Team, and the Team workspace must be available on its plan. There is no action to move a Team Project back to a personal workspace. ## How does editing work when several people open a Project? Nodespell lets one browser tab edit a Project at a time. Everyone else who opens it is viewing. The project bar shows who is editing. | Status | What it means | What you can do | | ------------------------ | ------------------------------------------------------------------------------ | ------------------------------------------------------------ | | **Editing** | You are editing. Teammates and your other tabs have to ask before taking over. | Keep working. | | **_name_ is editing** | A teammate is editing. | Choose **Request editing**. They are asked first. | | **Waiting…** | Your request was sent. | Wait, or **Cancel request**. | | **_name_ wants to edit** | A teammate asked to take over. | **Hand over** (your changes save first) or **Keep editing**. | | **Viewing** | Nobody, or another of your own tabs, is editing. | Select **Edit here** or click to start editing. | If you choose **Keep editing**, the teammate can ask again after a minute. While you are viewing, you can still browse Assets and inspect media, but you cannot change the graph, upload or start runs. Nodespell does not show other people's edits live inside your view while you are viewing. When you start editing, Nodespell first loads the latest saved Project. ## Who pays for runs in a Team Project? Runs in a Team Project are charged according to the Team's plan: - On a **Studio** Team, runs use the Team's shared credit pool. Your personal credits are not charged. - On a **Pro** Team, each member's runs use their own personal credits. See [Team credits and usage](https://docs.nodespell.com/teams/credits-and-usage/). ## Team Projects when membership changes - When a member is removed or leaves, the Team's Projects and Assets stay with the Team. - When the owner deletes the Team, all Team Projects become the owner's personal Projects. The owner must remove every other member first. ## Versions in Team Projects Saved versions work the same way as in personal Projects. Use **File** > **Save Version ...** while you are editing, and **File** > **Versions** to view, restore or delete them. Restoring a version replaces the live Project for everyone in the Team. See [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/). ## Frequently asked questions ### Can two team members edit the same Project at the same time? No. One tab edits at a time. Another member can request editing, and the current editor chooses whether to hand over. ### Can the Team owner see my drafts? No. Drafts in **My drafts** are visible only to the person who created them. ### Do I lose my changes if I hand over? No. When you choose **Hand over**, Nodespell saves your changes before the other person starts editing. ### Can any member archive or delete a shared Project? Yes. Any member who can open a shared Project can archive it, and delete it from the archive. Deletion cannot be undone. ## Related - [Members and roles](https://docs.nodespell.com/teams/members-and-roles/): what owners and members can do. - [Team credits and usage](https://docs.nodespell.com/teams/credits-and-usage/): who pays for runs. - [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/): editing in several tabs. - [Projects dashboard](https://docs.nodespell.com/projects/dashboard/): create and archive Projects. - [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/): saved versions. --- # Team credits, shared pools and usage > See which Nodespell credit balance pays for Team work, how a Studio shared pool works, and how owners track team credit usage, alerts and top-ups. Source: https://docs.nodespell.com/teams/credits-and-usage/ · Last updated: 2026-11-10 Team credits in Nodespell come from one of two places. On a Studio Team, work in the Team workspace uses one shared credit pool that the owner pays for. On a Pro Team, each member uses their own personal credits. Owners of a Studio Team can see where the pool's credits go, by day and by member, and turn on alerts before it runs out. ## Which balance pays for a run? | Where you run it | Balance used | | ------------------------------------ | ---------------------- | | A Project in your personal workspace | Your personal credits | | A Project in a Pro Team | Your personal credits | | A Project in a Studio Team | The Team's shared pool | Personal and Team credits are never combined. When a Studio pool is empty, your personal credits are not used for Team work, and when your personal credits run out, a Team's pool is not used for personal work. The credit display in the top bar names the balance in use, such as **Personal credits**, **Your credits in this team** or the Team's shared credit pool. ## How the Studio shared pool fills A Studio plan gives a number of monthly credits per team member, combined into one shared pool. The pool refreshes each billing cycle, and monthly credits do not roll over. The owner can also buy **Team pool top-ups**, which only that Team's Projects use. See [Billing](https://docs.nodespell.com/account/billing/). Removing a member frees their place but does not change the bill or the monthly credits unless the owner changes the plan. ## See where your team's credits go Studio Team owners have a full usage view. 1. Open **Account Settings** > **Teams**, or choose **Members** under **Team actions** in the Team workspace. 2. Select the Team, then open the **Credit usage** tab. 3. Choose a **Range**: **This cycle**, **Last cycle** or **Last 30 days**. | Part of the view | What it shows | | ---------------- | ---------------------------------------------------------------------------------------------------- | | Summary | **Used**, **Available now**, and when the next monthly credits arrive. | | Monthly meter | **Monthly credits used** against the cycle's included credits, as a percentage. | | Held credits | Credits held for work in progress, which may be released when it finishes. | | Pace estimate | Whether the monthly credits will last until they renew at the current pace. Top-ups are not counted. | | **Daily** chart | Credits used per UTC day. Lighter bars are partial days; click a bar for its breakdown. | | Member breakdown | Each member's credits and **Share**, including **Former member** and **Unattributed usage** rows. | **View as table** switches the chart to a table, and selecting a member filters the view. **Download CSV** exports the data. The view counts credits actually deducted from the shared pool, including top-ups; personal spending is excluded and days are in UTC. ## What members see Members of a Studio Team see their own usage in **Credit usage**: how much of the month's included credits the Team has used, **Team credits available now**, and how many credits they used this cycle. They do not see other members' usage. On a Pro Team there is no shared pool to track. Owners see **Each member uses their own credits**, and members do not get a **Credit usage** tab. ## Turn on pool alerts The owner of a Studio Team can turn on **Pool alerts** from the usage view. Alerts tell you when 80% of the month's included credits are used, and when no credits are available. Alerts are sent by email to your verified address and also appear in the app; without a verified email, they appear only in the app. To change them later, open **Teams** > **Credit usage** and choose **Alerts: On** or **Alerts: Off**. ## When the shared pool is empty When a Studio pool runs out, Team runs stop until credits are added. Members see that the shared pool is empty and can choose **Ask _owner_ to top up**. The owner gets the request in the app and by email. After asking, a member must wait before asking again. The owner can choose **Top up team credits**, which opens Team billing, or wait for the next monthly refresh. ## Frequently asked questions ### Can members top up a Studio Team's pool? No. Only the Team owner manages Team billing and top-ups. Members can ask the owner to top up from the empty-pool notice. ### Do my personal top-ups work in a Studio Team? No. Personal top-ups are added to your personal workspace only. Studio Team Projects use the Team's shared pool. ### Can members see how much each teammate used? No. Only the owner sees usage by member. Members see their own usage and the shared balance. ### Why does the usage view show held credits? Credits are held from the balance while work runs. You are charged only for completed work, and unused held credits are released. ### Are budgets or per-member limits available? The app does not offer per-member spending limits. Owners can use **Pool alerts** and the usage view to watch spending. ## Related - [Teams overview](https://docs.nodespell.com/teams/): Pro and Studio Teams. - [Members and roles](https://docs.nodespell.com/teams/members-and-roles/): who can manage billing. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): what credits are and what uses them. - [Billing](https://docs.nodespell.com/account/billing/): Team billing and top-ups. - [Team projects](https://docs.nodespell.com/teams/team-projects/): where Team runs happen. --- # Recipes — step-by-step Nodespell workflows > Nodespell recipes are complete, step-by-step workflows that chain several tools to reach one result, from product shots and talking videos to soundtracks. Source: https://docs.nodespell.com/recipes/ · Last updated: 2026-11-10 Nodespell recipes are end-to-end walkthroughs that combine several Creative Tools and nodes to reach one finished result, such as a relit product cutout, a talking video, or a scored and rendered clip. Each recipe starts from the outcome, lists what you need, shows the graph at a glance, and walks through every stage with the exact controls to use. ## Find a recipe by goal | Goal | What you start with | Tools used | Recipe | | ------------------------------------------------------ | ------------------------------------------- | -------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------- | | A finished, isolated product image | A written brief, optionally a product photo | Image Generator, Relight, Remove Background, Image Upscaler | [Product shot](https://docs.nodespell.com/recipes/product-shot/) | | A video shot from an approved still | One approved image | Image to Video, takes, Video Upscaler | [Image-to-video shot](https://docs.nodespell.com/recipes/image-to-video-shot/) | | A better version without losing earlier work | Any generated result | Image Editor, Variations, takes, model branches | [Revise without starting over](https://docs.nodespell.com/recipes/revise-without-starting-over/) | | A finished composite from existing images | A foreground and a background image | Image Compositor, Keyer, Color Correct, Transform, Merge, Crop | [Composite and finish](https://docs.nodespell.com/recipes/composite-and-finish/) | | The same character or product in many images and shots | One reference image | Image Generator with reference images, Variations, Image to Video, Reference Images to Video | [Consistent character](https://docs.nodespell.com/recipes/consistent-character/) | | A person speaking your script | A script and a portrait or clip | Prompt, Voice Generator, Speak | [Talking video](https://docs.nodespell.com/recipes/talking-video/) | | A refocused still with distance haze | One still image | Depth Anything V3 Image or Facet Depth V1, Image Compositor, ZDefocus, Depth Atmosphere | [Depth refocus and atmosphere](https://docs.nodespell.com/recipes/depth-refocus-and-atmosphere/) | | A clip with music and sound effects | A video clip | Music Generator, Sound Effects Generator, Timeline | [Soundtrack for a clip](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/) | | A longer shot joined with others | A video clip | Extend Video, takes, Timeline | [Extend a shot](https://docs.nodespell.com/recipes/extend-a-shot/) | | The best model for a prompt | One prompt | Prompt node, several model nodes, Design panel compare | [Compare models side by side](https://docs.nodespell.com/recipes/compare-models-side-by-side/) | ## How the recipes work Every recipe uses the same layout: - **What you need** lists the inputs, credits, and any browser requirements. - **Graph at a glance** shows the chain of nodes in one line, for example Prompt → Image Generator → Relight → Remove Background. - **Stages** are numbered steps with the exact labels you see in Nodespell. - **Check the result** lists what to inspect before you continue. - **Variations** shows two to four ways to adapt the recipe. Recipes often start in **Create**, where results appear under **Results**, and continue in **Graph**, where each node keeps its results as numbered takes. See [Workspace modes](https://docs.nodespell.com/start/workspace-modes/) and [Takes](https://docs.nodespell.com/graph/takes/). ## Where do recipes spend credits? Steps that run an AI model, such as Image Generator, Image to Video, or Voice Generator, use credits. Image-processing nodes such as Keyer, ZDefocus, and Merge, and arranging and rendering on the Timeline, do not generate new AI media. On paid plans, Nodespell shows a credit estimate in the Create composer and on the run button before you run, and existing takes are reused rather than regenerated unless you ask for new ones. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Frequently asked questions ### Do I need to finish a recipe in one session? No. Projects save automatically, and every node keeps its takes, so you can stop after any stage and continue later. ### Can I mix recipes? Yes. The recipes chain together: for example, make a [consistent character](https://docs.nodespell.com/recipes/consistent-character/), turn a still into a [video shot](https://docs.nodespell.com/recipes/image-to-video-shot/), [extend it](https://docs.nodespell.com/recipes/extend-a-shot/), then add a [soundtrack](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/). ### Are the models named in recipes required? No, unless the step says so. Model names are examples of current options; the model picker shows which models fit each tool. ### Can the Agent build a recipe for me? The in-product Agent can inspect your graph and propose or make visible changes to it. See [Agent](https://docs.nodespell.com/agent/) and [What the Agent can do](https://docs.nodespell.com/agent/what-agent-can-do/) for its current abilities and limits. ## Start here - [Product shot](https://docs.nodespell.com/recipes/product-shot/): Generate, relight, cut out, and finish a product image. - [Image-to-video shot](https://docs.nodespell.com/recipes/image-to-video-shot/): Turn an approved still into a video shot. - [Revise without starting over](https://docs.nodespell.com/recipes/revise-without-starting-over/): Edit, vary, and branch while keeping earlier work. - [Talking video](https://docs.nodespell.com/recipes/talking-video/): Script to voice to a speaking portrait. ## Related - [Make your first image](https://docs.nodespell.com/start/first-image/) — the shortest path to a first result - [Build your first graph](https://docs.nodespell.com/start/first-graph/) — connect nodes before you try a recipe - [Creative Tools](https://docs.nodespell.com/creative-tools/) — every tool used in these recipes - [Takes](https://docs.nodespell.com/graph/takes/) — choose which result each step uses - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) — how credits and estimates work --- # Make a product shot with relighting and a clean cutout > Generate a product still in Nodespell, relight it with studio presets, remove the background, then upscale and download the finished image. Source: https://docs.nodespell.com/recipes/product-shot/ · Last updated: 2026-11-10 The Nodespell product-shot recipe takes a written brief to a finished, isolated product image in four stages: generate the still with Image Generator, reshape the light with Relight, cut the product out with Remove Background, and enlarge and download the result. Every stage stays on the graph, so you can change one step without redoing the others. **What you need** - A short brief: the product, surface, camera angle, background, and mood. - Optional: a photo of the real product, uploaded to [Asset Manager](https://docs.nodespell.com/assets/asset-manager/), if the result must look like a specific object. - Credits. Image Generator, Relight, Remove Background, and Image Upscaler each run a model and use credits. **Graph at a glance:** Image Generator → Relight → Remove Background → Image Upscaler → download > **Where credits are spent:** Each of the four stages is a separate model run, and each run uses credits. On paid plans, the Create panel shows **Estimated per run** (or **Estimated total** when you ask for several generations), and the run button in Graph shows the estimate before you start. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ## Stage 1: Generate the base product image Use Create for the first image, because it gives you the guided controls and several candidates at once. 1. Open your Project and switch the workspace mode to **Create**. Create needs a window at least 1080 pixels wide. 2. Choose **Image Generator** from the Create tools rail. 3. Pick a model in the model picker. If the product must match a real object, choose a model that shows a **Reference images** field, then select **Choose assets**, pick the product photo, and select **Use selection**. 4. Write the brief in **Prompt**. Name the product, the surface it sits on, the camera angle, and the background. 5. Set **Aspect ratio** or the size control the model shows, and set **Generations** to between 1 and 4 to get several candidates. 6. Select **Generate**. Each finished image appears under **Results** on the Image Generator node. Step through them with **Show older result** and **Show newer result**, or open **View fullscreen** to check labels and edges. For the full Create walkthrough, see [Make your first image](https://docs.nodespell.com/start/first-image/) and [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/). ## Stage 2: Relight the chosen image Relight changes the lighting of an existing image while keeping the product recognizable. 1. Switch to **Graph** mode. The Image Generator node you used in Create is on the canvas. 2. Press `Tab`, type `Relight`, and add the **Relight** Creative Tool. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/). 3. Connect the Image Generator output to the Relight source image input. See [Connections](https://docs.nodespell.com/graph/connections/). 4. Select the Relight node. In **Parameters**, set **Relight method** to **Design lighting**. 5. Set **Lighting preset** to **Product softbox** as a starting point for a clean studio look. Adjust the **Key**, **Fill**, and **Rim** lights with **Direction**, **Elevation**, **Intensity**, and **Softness** only if the product needs a different shape or more separation. 6. Add **Lighting notes** for anything the controls do not cover, such as "keep the label readable". 7. With the Relight node selected, press `Cmd/Ctrl+Shift+R` or select the run button for the selection. Running the selected Relight node makes a new Relight take. Upstream nodes that already have a take, such as Image Generator, are reused rather than regenerated. Hover the run button to see which nodes will reuse a take and which will generate. See [Run nodes](https://docs.nodespell.com/graph/run-nodes/) and [Relight](https://docs.nodespell.com/creative-tools/relight/). > **Match reference needs unconnected inputs:** **Match reference** uses a second image as lighting guidance. It needs two local image slots, so Nodespell asks you to disconnect the source input before generating. Use the asset fields for both images instead of a connection when you choose this method. ## Stage 3: Remove the background 1. Press `Tab` and add **Remove Background**. 2. Connect the Relight output to the Remove Background source image input. 3. Select the Remove Background node and press `Cmd/Ctrl+Shift+R`. Remove Background isolates the subject from its surroundings. Skip this stage when the deliverable should keep the designed background and shadow from Relight. See [Remove Background](https://docs.nodespell.com/creative-tools/remove-background/). ## Stage 4: Upscale and download the finished image 1. Press `Tab` and add **Image Upscaler**. Connect the Remove Background output to it. 2. Run the Image Upscaler node. 3. Right-click the Image Upscaler node and choose **Download**, or choose **View Full Screen** to inspect the result first. See [Image Upscaler](https://docs.nodespell.com/creative-tools/image-upscaler/) and [Export](https://docs.nodespell.com/edit/export/). ## Check the result - Select the Relight node and the Remove Background node together. When exactly two image nodes are selected, the Design panel shows them as A and B in a comparison view, so you can check edges against the lit original. See [Viewers](https://docs.nodespell.com/edit/viewers/). - Zoom into thin parts, reflective surfaces, and text on the label. - Open the node's takes to make sure the take in use is the one you approved. Downstream nodes use the take each node is set to. See [Takes](https://docs.nodespell.com/graph/takes/). ## Variations - **Keep the studio background.** Stop after Relight and download that result. - **Match a real lighting setup.** Set **Relight method** to **Match reference** and choose a **Lighting reference** image whose light you want to copy. - **Try several lighting looks.** Add two or three Relight nodes to the same Image Generator output, give each a different **Lighting preset**, and compare them in the Design panel. - **Composite onto your own background.** Feed the cutout into an Image Compositor and merge it over a background image. See [Composite and finish existing assets](https://docs.nodespell.com/recipes/composite-and-finish/). ## Frequently asked questions ### Does rerunning Relight regenerate the base image? No. Running a downstream node reuses the takes its upstream nodes already have, so the Image Generator result is not regenerated or charged again. Choose **Generate** for a node only when you want a new take from it. ### Why is a Relight model missing when I choose Match reference? Only reference-capable models can match a lighting reference. If the current model cannot, Nodespell shows "Choose a reference-capable model to match a lighting reference." Pick another model in the model picker. ### Can I keep the product identical to my photo? Generated images can drift from the real object. Use a model with **Reference images**, keep the product photo connected, and compare each take against the photo before continuing. ### Where do the finished images go? Generated results appear in [Asset Manager](https://docs.nodespell.com/assets/asset-manager/) under the **Generated** view, as well as on their nodes, so you can reuse them in other Projects. ## Related - [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) — models, references, and output controls - [Relight](https://docs.nodespell.com/creative-tools/relight/) — every lighting control and preset - [Remove Background](https://docs.nodespell.com/creative-tools/remove-background/) — cutout models and limits - [Takes](https://docs.nodespell.com/graph/takes/) — choose which result each stage uses - [Revise without starting over](https://docs.nodespell.com/recipes/revise-without-starting-over/) — change one stage and keep the rest --- # Turn an approved still into a video shot > Animate an approved still in Nodespell with Image to Video, compare several video takes, choose the best one, then upscale and download the shot. Source: https://docs.nodespell.com/recipes/image-to-video-shot/ · Last updated: 2026-11-10 The Nodespell image-to-video recipe turns one approved still into a finished video shot. You hand the exact image you approved to the Image to Video Creative Tool as the start frame, describe the motion, generate a few takes, keep the best one, and optionally upscale it before download. The still stays untouched, so you can try new motion without losing it. **What you need** - An approved still: a result on an image node in your Project, or an image in [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). - A motion brief: what the subject does, how the camera moves, and what must stay unchanged. - Credits. Each Image to Video generation uses credits. Video Upscaler, if you use it, is a separate run. **Graph at a glance:** Image node (or Asset) → Asset node "Selected historical result" → Image to Video → Video Upscaler → download > **Where credits are spent:** Credits are spent when you select **Generate** for Image to Video and when you run Video Upscaler. On paid plans, Create shows **Estimated per run** or **Estimated total** for the number of generations before you select **Generate**. For many video models, longer durations and higher resolutions raise the estimate. ## Stage 1: Send the approved still to Image to Video The most reliable path starts from the result you approved, because Nodespell then uses that exact image rather than whatever the node outputs later. 1. In **Create**, open the tool that made the still (for example **Image Generator**) and select its node. 2. Under **Results**, step to the still you approved with **Show older result** and **Show newer result**. 3. Select **Continue with this result**. 4. In the **Continue with this result** dialog, choose **Image to Video** ("Use this image as the starting frame for a video"). Nodespell keeps that exact output as a visible Asset node named **Selected historical result** and connects it to the new Image to Video draft as the **Start frame**. The composer shows "This tool will use the exact selected result through a visible Asset node." To start from an uploaded image instead, choose **Image to Video** in the Create tools rail, select **Choose assets** in **Start frame**, pick the image, and select **Use selection**. ## Stage 2: Describe the motion and generate takes 1. Pick a model in the model picker. The recommended models appear first; see [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). 2. Write the motion brief in **Prompt**. Name the action, the camera move, and the pace, and say what must not change. For example: "The bottle turns slowly to the left while the camera pushes in. The label and background stay unchanged." 3. If the model shows an end-frame field, add a second still to guide where the shot ends. 4. Set **Duration**, **Resolution**, and **Aspect ratio** if the model shows them. Turn **Generate audio** on or off if the model offers it. 5. Set **Generations** to 2–4 to get several takes to choose from. Use 1 while you are still testing the brief. 6. Select **Generate**. The batch progress shows each generation as **Queued**, **Running**, **Complete**, or **Failed**. Successful generations are kept even if another in the batch fails. See [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/). ## Stage 3: Compare takes and choose one 1. Step through the clips in **Results**, or open **View fullscreen** to judge motion at full size. 2. Switch to **Graph** and select the Image to Video node. Open its takes from **Takes** next to the run button, or right-click the node and choose **Takes…**. Video takes show a poster frame until you select **Play**. 3. Preview a take, then select **Use Take** with its number to make it the node's output. To stay on it even after new runs, open **More take actions** and choose **Keep this take**. Downstream nodes always use the take the node is set to. See [Takes](https://docs.nodespell.com/graph/takes/). ## Stage 4: Upscale and download 1. In Graph, press `Tab`, add **Video Upscaler**, and connect the Image to Video output to it. 2. Select the Video Upscaler node and press `Cmd/Ctrl+Shift+R`. 3. Right-click the finished node and choose **Download**. Skip the upscale if the shot is already large enough. See [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/) and [Export](https://docs.nodespell.com/edit/export/). ## Check the result - Play the chosen take at full size and look at the first frames: the subject and composition should match the still. - Check hands, faces, text, and product edges, which drift most in generated motion. - Confirm the take in use on the Image to Video node is the one you approved before you run anything downstream. ## Variations - **Guide the ending.** Add an end frame (for example, a reframed version of the still made with [Variations](https://docs.nodespell.com/creative-tools/variations/)) when the shot must land on a specific composition. - **Several references instead of one start frame.** Use [Reference Images to Video](https://docs.nodespell.com/creative-tools/reference-images-to-video/) when a character, style, or location should come from more than one image. - **Make it longer.** Continue the chosen take with [Extend Video](https://docs.nodespell.com/recipes/extend-a-shot/). - **Add sound.** Score the clip with music and effects in [Add a soundtrack to a clip](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/). ## Frequently asked questions ### Why did Nodespell add an Asset node when I continued? **Continue with this result** preserves the exact image you picked as a visible Asset node, named **Selected historical result**, so the video always starts from that still even if the original node gets new takes later. ### Which tools can I continue to from an image result? The **Continue with this result** dialog offers **Image Upscaler** and **Image to Video**. For other tools, connect the image node in Graph or choose the image as an asset. ### Can I change the motion without regenerating the still? Yes. Edit the **Prompt** or settings on the Image to Video node and generate again. The still and its node are not rerun. ### Does a failed generation use credits? Each take records its charge. Open the take's **Details** to see whether it was charged, refunded, or not charged. ## Related - [Make your first image-to-video clip](https://docs.nodespell.com/start/first-image-to-video/) — the beginner version of this recipe - [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) — models, frames, and motion controls - [Takes](https://docs.nodespell.com/graph/takes/) — preview, use, and keep takes - [Extend a shot](https://docs.nodespell.com/recipes/extend-a-shot/) — continue the chosen clip - [Video Upscaler](https://docs.nodespell.com/creative-tools/video-upscaler/) — enlarge the finished shot --- # Keep a character or product consistent across shots > Build one reference image in Nodespell and feed it to Image Generator, Variations, and the video tools to keep a character or product consistent. Source: https://docs.nodespell.com/recipes/consistent-character/ · Last updated: 2026-11-10 Keeping a character or product consistent in Nodespell means every image and shot starts from the same reference image instead of a fresh prompt. You make or upload one approved reference, keep it as an Asset node, and connect that one node to every tool that accepts reference images. New angles come from Variations, new scenes from reference-capable image models, and motion from the video tools. **What you need** - One clear image of the subject: a front view with even light and a plain background works best. Upload it to [Asset Manager](https://docs.nodespell.com/assets/asset-manager/), or generate it in stage 1. - A short description of the subject you reuse word for word in every prompt, for example "a red-haired courier in a yellow raincoat with a silver backpack". - Credits for each generation. Reusing the same reference costs nothing extra. **Graph at a glance:** Reference Asset → Variations (angles) · Reference Asset → Image Generator with **Reference images** (new scenes) · chosen still → Image to Video · Reference Asset + stills → Reference Images to Video > **Where credits are spent:** Every Image Generator, Variations, Image to Video, and Reference Images to Video run uses credits. On paid plans, Create shows the estimate before you select **Generate**, and the run button in Graph shows it before you run. ## Stage 1: Make the reference image 1. In **Create**, choose **Image Generator**. 2. In **Prompt**, describe the subject on a plain background, in full view, with even light. Include the reusable subject description. 3. Set **Generations** to 4 and select **Generate**. 4. In **Results**, pick the clearest version. 5. Switch to **Graph**, select the Image Generator node, open **Takes**, and select **Use Take** for the chosen take. Then open **More take actions** and choose **Keep this take**, so later runs never replace it. If you already have a photo or illustration, skip generation: press `Tab` in Graph, add an **Asset** node, right-click it, choose **Load Asset**, and pick the image. ## Stage 2: Make new angles with Variations Variations keeps the source recognizable while changing the camera. 1. Press `Tab` and add **Variations**. Connect the reference to it, or set **Source image** with **Choose assets** in Create. 2. Set **Variation mode** to **Reframe** and choose a **Shot type**, such as **Profile view**, **Back view**, **Three-quarter view**, or **Close-up**. 3. For a change of pose or outfit, use **Custom** and write **Variation instructions** that say what changes and what stays: "Same courier and raincoat, now sitting on a bench. Keep the face, hair, and colors identical." 4. Run the node. Repeat with other shot types to build a small turnaround set. See [Variations](https://docs.nodespell.com/creative-tools/variations/). ## Stage 3: Put the subject in new scenes 1. Press `Tab` and add **Image Generator**. 2. Select it and choose a model that shows a **Reference images** field. Not every model accepts references; if the field is missing, switch models in the model picker. 3. Connect the reference (and, optionally, one or two of the angle variations) to the reference images input. One output can feed any number of nodes. 4. In **Prompt**, write the new scene and repeat the subject description exactly. 5. Generate several takes and keep the one that matches best. For a small change to an existing scene, such as swapping a prop, use [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) on that scene instead of generating a new one. ## Stage 4: Bring the subject into motion There are two video paths, and both accept your reference: | Tool | Use it when | What you give it | | ----------------------------------------------------------------------- | ------------------------------------------------------------------- | ------------------------------------------------------------ | | [Image to Video](https://docs.nodespell.com/creative-tools/image-to-video/) | One approved still should become the first frame of the shot. | A **Start frame**, and an end frame if the model offers one. | | [Reference Images to Video](https://docs.nodespell.com/creative-tools/reference-images-to-video/) | The shot should be new, but the subject must match your references. | Several **Reference images**, plus a prompt. | 1. For a reference-led shot, choose **Reference Images to Video** in Create, or add it in Graph. 2. Add the reference and the best angle variations under **Reference images**. 3. Write the prompt. Some models, such as Seedance 2.5 Reference to Video and Kling 4.0 Reference to Video, let you point to each image in the prompt as `@Image1`, `@Image2`, and so on. Check the model's node description for how it numbers references. 4. Generate, compare takes, and keep the best. ## Check the result - Select the reference and a new image together; the Design panel shows two selected images as A and B. Check the face, hair, colors, logo, and proportions. - Line up all the takes in the node's takes and reject any that drift, rather than correcting them later. - In video, scrub to the middle and end of the clip, where identity drift usually appears. ## Variations - **Products instead of people.** Use a clean product photo as the reference and include the brand name and materials in every prompt. Pair with the [product shot recipe](https://docs.nodespell.com/recipes/product-shot/). - **Repeatable results.** Many models show **Seed** under **More settings**. Keep the same seed while you change only the prompt to reduce unrelated changes. - **A talking character.** Feed a consistent portrait into Speak; see [Make a talking video](https://docs.nodespell.com/recipes/talking-video/). - **One shared prompt.** Put the subject description in a **Prompt** node and connect it to several model nodes, so a wording change updates every shot. See [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/). ## Frequently asked questions ### Which Nodespell tools accept reference images? Image Generator (with models that show **Reference images**), Image Editor, Variations, Relight, Image to Video (start and end frames), Reference Images to Video, and Speak all take image inputs. The exact fields depend on the model you choose. ### Why does my character still change between shots? Generated models do not guarantee identity. Use more than one reference angle, keep the subject description identical in every prompt, and keep only the takes that match. ### How do I refer to a specific reference image in the prompt? Models that support it describe the syntax in their node description; Seedance 2.5 Reference to Video and Kling 4.0 Reference to Video use `@Image1`, `@Image2`, and so on. Check the model's description in the model picker or the Library panel. ### Can I reuse the reference in another Project? Yes. Generated images appear in [Asset Manager](https://docs.nodespell.com/assets/asset-manager/) under **Generated**, and uploaded images are in your library, so you can load the same reference in any Project. ## Related - [Variations](https://docs.nodespell.com/creative-tools/variations/) — reframes and directed variations - [Reference Images to Video](https://docs.nodespell.com/creative-tools/reference-images-to-video/) — reference-led video shots - [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) — models with reference images - [Takes](https://docs.nodespell.com/graph/takes/) — keep the reference take fixed - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — find reference-capable models --- # Make a talking video from a script and a portrait > Turn a written script into speech with Nodespell Voice Generator, then make a portrait or clip speak it with the Speak Creative Tool. Source: https://docs.nodespell.com/recipes/talking-video/ · Last updated: 2026-11-10 The Nodespell talking-video recipe turns a script and a portrait into a video of that person speaking. Voice Generator reads the script aloud as an audio take, and the Speak Creative Tool animates a still image, or lip-syncs an existing clip, to that audio. Because the voice is its own node, you can change the delivery without regenerating the face, and the other way round. **What you need** - A script, written the way it should be spoken. - A portrait: a clear, front-facing image of the face, uploaded to [Asset Manager](https://docs.nodespell.com/assets/asset-manager/) or generated in Nodespell. For lip-sync, an existing video clip of the person instead. - Credits for two runs: Voice Generator and Speak. **Graph at a glance:** Prompt (script) → Voice Generator → Speak ← Portrait Asset > **Where credits are spent:** Voice Generator and Speak each run a model and use credits. On paid plans, Create shows **Estimated per run** before you select **Generate**, and the run button in Graph shows the estimate for the selection. Rerunning Speak reuses the existing voice take, so the voice is not charged again. ## Stage 1: Write the script in a Prompt node 1. Open the Project in **Graph** mode. 2. Press `Tab` and add a **Prompt** node. 3. Double-click the Prompt node and type the script exactly as it should be spoken. Keeping the script in its own node lets you reuse it and edit it in one place. See [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/). ## Stage 2: Generate the voice 1. Press `Tab` and add **Voice Generator**. 2. Connect the Prompt output to the Voice Generator text input (shown as **Text** on ElevenLabs TTS V3). 3. Select the Voice Generator node. In **Parameters**, choose a voice. With ElevenLabs TTS V3, for example, choose a **Voice** and adjust **Stability**, **Style**, and **Speed**. Lower stability sounds more varied; higher stability is steadier. 4. Press `Cmd/Ctrl+Shift+R` to run it, or open **Run options** and choose a batch such as **Generate 3 takes** to hear several reads. 5. Play each take, then select **Use Take** for the best read and choose **Keep this take** from **More take actions**. You can also do this in **Create**: choose **Voice Generator**, type the script into the model's text field, and select **Generate**. See [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/). ## Stage 3: Make the portrait speak 1. Press `Tab` and add an **Asset** node. Right-click it, choose **Load Asset**, and pick the portrait. 2. Press `Tab` and add **Speak**. 3. Connect the portrait to the Speak image input (shown as **Source image** in the Speak controls) and the Voice Generator output to the Speak audio input. 4. Select the Speak node. Set **Resolution**, and adjust the video prompt if the model offers one (P-Video Avatar, for example, starts with "The person is talking."). 5. Run the Speak node. Voice Generator already has a take, so only Speak generates. See [Speak](https://docs.nodespell.com/creative-tools/speak/). ## Which Speak model fits your source? Speak groups several models that need different inputs. The model picker shows only models that fit the tool. | Starting point | Example model | Inputs it needs | | ----------------------------------------------------------- | ------------------------------ | --------------------------------------------------------------------------------- | | A still portrait and your own audio | P-Video Avatar, Omni Human | An image and an audio track | | A still portrait and a typed script, no separate voice step | P-Video Avatar | An image and a **Voice script**, with a built-in **Voice** and **Voice Language** | | An existing video of the person | Kling Lip Sync, Sync Lipsync 2 | A video and an audio track | With P-Video Avatar you can skip Voice Generator entirely: leave **Audio** empty and type the script into **Voice script**. The node then needs either audio or a voice script before it can run. ## Check the result - Play the Speak take with sound and watch the lips on hard consonants such as "p", "b", and "m". - Check the first and last second: the face should match the portrait and the clip should not cut off a word. - If the timing is wrong, change the voice take first; it is cheaper than regenerating video and the face can stay the same. ## Variations - **Lip-sync existing footage.** Switch the Speak node to a lip-sync model such as Sync Lipsync 2 and connect a video instead of an image. Changing the model of a node that has run creates a branch, so the first version stays. - **A consistent presenter.** Build the portrait with the [consistent character recipe](https://docs.nodespell.com/recipes/consistent-character/) so the same face can speak in several clips. - **Longer scripts.** Split the script into several Voice Generator nodes. ElevenLabs TTS V3 accepts previous and next text to keep the delivery consistent across pieces. Then join the Speak clips on the Timeline; see [Add a soundtrack to a clip](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/) for the Timeline steps. - **Add music under the speech.** Place the talking clip and a Music Generator take on the Timeline and lower the music **Gain**. ## Frequently asked questions ### Do I need Voice Generator to use Speak? No. Some Speak models, such as P-Video Avatar, accept a typed **Voice script** and a built-in voice. Use Voice Generator when you want more control over the voice or want to reuse the same audio. ### Why is the Speak node not ready to run? Each Speak model needs specific inputs. P-Video Avatar needs a source image plus either audio or a voice script; lip-sync models need both a video and an audio track. ### Can I change the voice without regenerating the face? The face is regenerated whenever Speak runs, because the mouth must follow the new audio. You can change the voice take freely, but you need a new Speak run to sync to it. ### Which languages does the voice support? It depends on the model. Check the voice and language options shown in **Parameters** for the model you picked, or its description in the model picker. ## Related - [Voice Generator](https://docs.nodespell.com/creative-tools/voice-generator/) — voices and delivery controls - [Speak](https://docs.nodespell.com/creative-tools/speak/) — talking-avatar and lip-sync models - [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/) — keep the script in a Prompt node - [Takes](https://docs.nodespell.com/graph/takes/) — choose the best read and the best clip - [Timeline](https://docs.nodespell.com/edit/timeline/) — join several talking clips --- # Extend a video shot and join it on the Timeline > Continue a video clip in Nodespell with Extend Video, compare several extended takes, keep the best, and join the shots on the Timeline into one file. Source: https://docs.nodespell.com/recipes/extend-a-shot/ · Last updated: 2026-11-10 The Nodespell extend-a-shot recipe makes a clip longer without regenerating it from scratch. The Extend Video Creative Tool continues a source clip after its end, before its start, or in both directions, guided by an optional prompt. You generate a few extended versions, keep the best, and then join it with neighboring shots on the Timeline and render one file. **What you need** - A source clip: a video result on a node in your Project, or a video in [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). - A short note of what should happen next, or nothing if a natural continuation is enough. - Credits for each extension. For some models, both the source seconds and the output seconds count toward the cost. **Graph at a glance:** Source video → Extend Video → (other shots) → Timeline → render → download > **Where credits are spent:** Each Extend Video generation uses credits. On paid plans, Create shows **Estimated per run** (or **Estimated total** for several generations) before you select **Extend**, and the Graph run button shows the estimate before you run. Arranging and rendering on the Timeline does not generate new AI media. ## Stage 1: Set up Extend Video 1. In **Create**, choose **Extend Video** ("Continue a source clip before, after, or in both directions."). 2. Pick a model. For example, Wan 3.0 Extend Video is the recommended model; Wan 3.0 Prime Extend Video offers faster generation with the same controls. 3. In **Source video**, select **Choose assets**, pick the clip (generated clips are under **Generated**), and select **Use selection**. In Graph you can connect the clip's output to the Extend Video source input instead. 4. Choose the extension direction. With Wan 3.0 Extend Video, **Extension Direction** offers **Add to End**, **Add to Start**, and **Both Ends**. 5. In **Prompt**, describe what happens in the new footage, or leave it empty for a natural continuation. 6. Set **Total Duration** and **Resolution**. Turn **Generate Audio** on if you want sound in the new footage. See [Extend Video](https://docs.nodespell.com/creative-tools/extend-video/). ## Know your model's limits Extend models differ in what they return and how long they can run. These rules come from the node descriptions in Nodespell: | Model | What the result contains | Length rules | | --------------------------------------------------- | ----------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------ | | Wan 3.0 Extend Video and Wan 3.0 Prime Extend Video | The full extended clip, not only the new footage. Framing follows the source. | **Total Duration** must be longer than the source; source plus output cannot exceed 30 seconds; **Auto** needs a source shorter than 15 seconds. | | FLUX.3 Extend Video | Continues an MP4 beyond its final frames. | Source under 50 MB and 15 seconds; continuation of 5–20 seconds at up to 1080p. | For other extend models, read the model's description in the model picker before you plan the edit. ## Stage 2: Generate several extensions and choose one 1. Set **Generations** to 2–4. 2. Select **Extend**. 3. Watch each result under **Results**, using **Show older result**, **Show newer result**, and **View fullscreen**. Look closely at the seam where the source ends and the new footage begins. 4. Switch to **Graph** and select the Extend Video node. Open **Takes** next to the run button, or right-click the node and choose **Takes…**. 5. Preview each take, select **Use Take** for the best one, then open **More take actions** and choose **Keep this take** so later runs don't replace it. See [Takes](https://docs.nodespell.com/graph/takes/). ## Stage 3: Join the shots on the Timeline How you join depends on what the model returned. Wan 3.0 Extend Video already contains the source, so place only the extended clip. If your model returns only the new footage, place the source first and the extension directly after it. 1. Press `Tab` and add a **Timeline** node. 2. Connect the clips to the Timeline **Assets** input in the order you want them: the extended shot, then any following shots. 3. Select the Timeline node to open its editor in the Design panel. 4. Open **Media** and select **Add all to timeline**, or drag each clip onto the video track. 5. Turn on **Snap** and butt the clips together. Use the **Trim** tool to remove repeated frames at a join, and **Close gap** to remove empty space. 6. Play across each cut and adjust. See [Timeline](https://docs.nodespell.com/edit/timeline/). ## Stage 4: Render and download 1. Open **Export**. 2. Set **Range** to **Whole timeline** and keep **Include audio** on if the clips have sound. 3. Select **Render**, then **Download** when it finishes. The render also feeds the Timeline node's **Video** output for further graph work. See [Export](https://docs.nodespell.com/edit/export/). ## Check the result - Play the join at full size. Watch for a jump in color, framing, or motion where the extension starts. - Check that the subject stays the same in the new footage; extension models do not guarantee exact preservation. - If you extended in both directions, check both seams. ## Variations - **Join without the Timeline.** In Graph, connect clips to a **Join Video** node, which joins connected videos end to end in connection order. Set its **Resolution** rule and **Fit mode** if the clips differ in size. - **Trim before joining.** Add **Trim Video** to cut the start or end of a clip by seconds or frames before it reaches the Timeline. - **Make a long shot in steps.** Extend the chosen take again with a new Extend Video node, keeping each step as its own node and take. - **Add sound afterwards.** Score the finished sequence with [Add music and sound effects to a video clip](https://docs.nodespell.com/recipes/soundtrack-for-a-clip/). ## Frequently asked questions ### Why is my extended clip the same length as the source plus the new part? Some models, such as Wan 3.0 Extend Video, return the full extended clip rather than only the new footage. Do not place the source again before it on the Timeline. ### Can I extend a clip backward? Yes, with models that support it. Wan 3.0 Extend Video offers **Add to Start** and **Both Ends** under **Extension Direction**. ### Why was my Total Duration rejected? For Wan 3.0 Extend Video, **Total Duration** must be longer than the source, and source plus output cannot exceed 30 seconds. **Auto** works only with a source shorter than 15 seconds. ### Does extending change my original clip? No. Extend Video is a new node with its own takes; the source clip and its node are unchanged. ## Related - [Extend Video](https://docs.nodespell.com/creative-tools/extend-video/) — models and controls - [Takes](https://docs.nodespell.com/graph/takes/) — compare and keep extended versions - [Timeline](https://docs.nodespell.com/edit/timeline/) — cuts, trims, and rendering - [Video editing](https://docs.nodespell.com/edit/video-editing/) — Join Video and Trim Video - [Turn an approved still into a video shot](https://docs.nodespell.com/recipes/image-to-video-shot/) — make the first shot --- # Add music and sound effects to a video clip > Generate music and sound effects in Nodespell, place them under a video clip on the Timeline, balance the mix, and render one finished video file. Source: https://docs.nodespell.com/recipes/soundtrack-for-a-clip/ · Last updated: 2026-11-10 The Nodespell soundtrack recipe scores a silent or rough clip. You generate a music bed with Music Generator and individual sounds with Sound Effects Generator, connect them and the clip to a Timeline node, line them up on separate tracks, balance the levels, and render one video with the mixed audio. The music, effects, and picture stay separate nodes, so you can replace any one of them later. **What you need** - A video clip: a generated shot (for example from [Turn an approved still into a video shot](https://docs.nodespell.com/recipes/image-to-video-shot/)) or an uploaded video. - A note of the clip's length and the moments that need a sound, such as a door closing at 3 seconds. - Credits for each music and sound-effect generation. - A current Chromium- or Firefox-based desktop browser; the Timeline editor needs one. **Graph at a glance:** Video clip + Music Generator + Sound Effects Generator (one per sound) → Timeline → render → download > **Where credits are spent:** Music Generator and each Sound Effects Generator run use credits. On paid plans, Create shows **Estimated per run** before you select **Generate**, and the Graph run button shows the estimate for the selection. Arranging and rendering on the Timeline does not generate new AI media. ## Stage 1: Generate the music bed 1. In **Create**, choose **Music Generator**. 2. Pick a model. For example, Google Lyria 3 Pro is built for longer, structured tracks. 3. In **Prompt**, describe genre, instruments, mood, and tempo. With Lyria 3 Pro you can add section tags such as `[Intro]`, `[Verse]`, `[Chorus]`, and `[Outro]` to guide the arrangement; they are not exact timeline markers. 4. Set **Generations** to 2 or 3 and select **Generate**. 5. Listen to each result and keep the best one. See [Music Generator](https://docs.nodespell.com/creative-tools/music-generator/). ## Stage 2: Generate each sound effect Make one sound per generation. Short, specific prompts work better than one prompt for a whole sequence. 1. In **Create**, choose **Sound Effects Generator**. 2. Describe one sound, for example "heavy wooden door closing in a large hall". 3. Set the length and loop options the model shows. ElevenLabs Sound Effects V2, for example, has a duration setting and a **Loop** switch for ambient beds. 4. Select **Generate**. Repeat for every sound you need. See [Sound Effects Generator](https://docs.nodespell.com/creative-tools/sound-effects-generator/). ## Stage 3: Collect everything in a Timeline 1. Switch to **Graph** mode. 2. Press `Tab` and add a **Timeline** node. 3. Connect the video clip, the Music Generator node, and each Sound Effects Generator node to the Timeline **Assets** input. It accepts many connections. 4. Select the Timeline node. The Timeline editor opens in the Design panel. See [Timeline](https://docs.nodespell.com/edit/timeline/). ## Stage 4: Arrange the tracks 1. Open **Media**. Everything connected to the Assets input is listed there. 2. Select **Add all to timeline**, or drag each item onto a track. Use **Add to timeline** on a single item to place just that one. 3. If there is no free audio track, select **Add track** and add an audio track, so music and effects sit on separate tracks. 4. Drag each sound effect to its moment. Turn on **Snap** to align clip edges, and use `J`, `K`, and `L` to shuttle playback while the timeline has focus. 5. Trim the music to the clip length with the **Trim** tool, or split it with **Split at playhead**. ## Stage 5: Balance the mix 1. Select the music clip and open **Inspector**. Lower **Gain** so it sits under the effects, and set **Fade In** and **Fade Out**. 2. Select each effect clip and adjust its **Gain**. 3. If the video has its own generated audio you do not want, use **Mute track** on that track, or select the clip and choose **Detach audio** to work with it separately. 4. Open **Mixer** to compare track levels while you play the timeline. ## Stage 6: Render and download 1. Open **Export**. 2. Set **Range** to **Whole timeline**, and make sure **Include audio** is on. 3. Choose a preset, or set **Format**, **Resolution**, and **Frame rate**. 4. Select **Render**. When it finishes, select **Download**. The render also "Feeds the Video output of this node", so you can connect the finished video to other nodes. The render status reads **Out of date** after you change the timeline; select **Render again** to update it. See [Export](https://docs.nodespell.com/edit/export/). ## Check the result - Play the rendered file outside the editor at normal volume. Effects should be clear over the music. - Scrub to each effect and check it lands on the right frame. - Check the last second: the music should fade, not cut. ## Variations - **Score to the picture.** In Graph, model nodes such as Sonilo V1.1 Video to Music take the video as input and generate music that follows its pacing. Connect the clip, run the node, and use its audio on the Timeline. - **Sound design from the video.** Mirelo Video To SFX and MMAudio V2 generate sounds that follow the footage. Use them when timing should follow on-screen action. - **Several shots.** Put several clips on the video track before adding music; see [Extend a shot](https://docs.nodespell.com/recipes/extend-a-shot/). - **Dialogue too.** Add a Voice Generator take on its own audio track; see [Make a talking video](https://docs.nodespell.com/recipes/talking-video/). ## Frequently asked questions ### Does rendering the Timeline use credits? Rendering assembles media you already have and does not generate new AI media. Credits are spent when you generate the music and sound effects. ### Why doesn't my audio show up in Media? Media lists what is connected to the Timeline **Assets** input. Connect the Music Generator or Sound Effects Generator node to that input, and make sure the node has a take. ### Can I change the music after rendering? Yes. Generate a new music take, choose it on the Music Generator node, then select **Render again** on the Timeline. The render status shows **Out of date** until you do. ### Why is there no audio in my export? Check that **Include audio** is on in **Export** and that no audio track is muted. ## Related - [Timeline](https://docs.nodespell.com/edit/timeline/) — every Timeline tool, track control, and shortcut - [Music Generator](https://docs.nodespell.com/creative-tools/music-generator/) — music models and prompts - [Sound Effects Generator](https://docs.nodespell.com/creative-tools/sound-effects-generator/) — single-sound generation - [Video editing](https://docs.nodespell.com/edit/video-editing/) — video nodes outside the Timeline - [Export](https://docs.nodespell.com/edit/export/) — render settings and downloads --- # Refocus an image and add depth haze from a depth map > Generate a depth map in Nodespell, then use ZDefocus and Depth Atmosphere in an Image Compositor to refocus a still and add distance haze. Source: https://docs.nodespell.com/recipes/depth-refocus-and-atmosphere/ · Last updated: 2026-11-10 The Nodespell depth recipe turns a flat still into one with a chosen focus plane and distance haze. A depth model node creates a grayscale depth map of the image, then inside an Image Compositor, ZDefocus blurs everything off the focus plane and Depth Atmosphere adds fog, color, and bloom that grow with distance. Only the depth map uses credits; the compositor work runs in the editor and can be adjusted freely. **What you need** - A still image, as an Asset or a result on a node in your Project. - Credits for one depth-map run. - Familiarity with compositors helps. Depth maps here work like a Z pass in other compositing tools: one channel, near and far encoded as brightness. **Graph at a glance:** Image → Depth Anything V3 Image (or Facet Depth V1) → Image Compositor { Compositor Input (image) + Compositor Input (depth) → ZDefocus → Depth Atmosphere → Compositor Output } > **Where credits are spent:** The depth model node is the only paid step. On paid plans, the run button shows its estimate before you run. ZDefocus and Depth Atmosphere are image-processing nodes and do not use credits, so you can tune them as often as you like. ## Stage 1: Generate the depth map 1. Open the Project in **Graph** mode. If the image is not on the canvas, press `Tab`, add an **Asset** node, right-click it, choose **Load Asset**, and pick the image. 2. Press `Tab` and add a depth model node, for example **Depth Anything V3 Image** or **Facet Depth V1**. 3. Connect the image to the depth node's **Image** input. 4. Select the depth node and press `Cmd/Ctrl+Shift+R`. Both nodes output a grayscale map with near surfaces in white and distant surfaces in black. Depth is relative, not a distance in meters. | Depth node | Useful settings | Outputs | | ----------------------- | ---------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------- | | Depth Anything V3 Image | **Output format**: keep **PNG** to avoid lossy compression. | One grayscale depth image. | | Facet Depth V1 | **Ignore transparent objects** (**Standard** or **See-through**), **Processing resolution** (**1K**, **1.5K**, **2K**), **Scale**. | **Depth map (PNG)**, plus **Depth (EXR)** and **Raw depth (NPY)** with floating-point values. | See [Depth and defocus](https://docs.nodespell.com/edit/depth-and-defocus/) for the full list of depth tools. ## Stage 2: Bring image and depth into a compositor 1. Press `Tab` and add **Image Compositor**. 2. Connect the image to the Image Compositor **Images** input first, then connect the depth map. The first connection is index 0, the second is index 1. 3. Select the Image Compositor and press `Enter` (or right-click and choose **Enter Node**). 4. Leave the existing **Compositor Input** on **Index** `0`; this is the image. 5. Press `Tab`, add a second **Compositor Input**, and set **Index** to `1`; this is the depth map. ZDefocus works only inside a compositor. See [Compositing](https://docs.nodespell.com/edit/compositing/). ## Stage 3: Refocus with ZDefocus 1. Press `Tab` and add **ZDefocus**. 2. Connect the image Compositor Input to the ZDefocus image input and the depth Compositor Input to its depth input. 3. Check **Depth Encoding** is **White Near**, which matches both depth nodes. Use **Black Near** only for maps where near is dark. 4. Set **Output** to **False Colour Focus** to see which areas are in focus. 5. Adjust **Focus Depth** (the normalized depth that stays sharp) until your subject is in focus, and **Focus Range** to widen or narrow the sharp zone. 6. Set **Near Blur** and **Far Blur** for the strength in front of and behind the focus plane, and **Max Radius** for the largest bokeh in pixels. 7. Choose an **Aperture**: **Disc**, **Blades**, **Heart**, **Star**, **Ring**, or **Image** (uses a picture connected to **Kernel**). Raise **Highlight Gain** to make bright points bloom into bokeh shapes. 8. Set **Output** back to **Result**. Switch **Quality** to **High quality** for the final frame. Other **Output** views help you debug: **Depth View** shows the depth as read, **Blur View** shows blur amount, **Kernel** shows the aperture shape, and **Layer Setup** shows how depth is split into layers. ## Stage 4: Add distance haze with Depth Atmosphere 1. Press `Tab` and add **Depth Atmosphere**. 2. Connect the ZDefocus output to **Image** and the depth Compositor Input to **Depth**. 3. Set **Start** and **End**, the normalized distances where atmosphere begins and reaches full strength. 4. Set **Air Density** for how thick the fog is, **Maximum Opacity** to cap the haze, and **Air Colour** to tint it. 5. Choose **Falloff**: **Exponential**, **Linear**, or **Custom** to draw your own **Falloff Curve**. 6. Optionally, in the grade and diffusion controls, use **Exposure**, **Contrast**, and **Saturation** to change distant tones, **Image Diffusion** to soften distant detail, and **Highlight Bloom** to spread light from bright areas. 7. Use **Output** views such as **Fog Mask**, **Distance Influence**, or **Atmosphere Only** to check coverage, then return to **Result**. ## Stage 5: Output and download 1. Connect the Depth Atmosphere output to **Compositor Output**, replacing the default connection. 2. Press `U` to return to the parent graph. 3. Select the Image Compositor to see the result in the Design panel, then right-click it and choose **Download**. ## Check the result - Look at subject edges. ZDefocus spreads foreground blur past object edges; turn on **Edge Anti-aliasing** if edges look stepped. - Set Depth Atmosphere **Output** to **Fog Mask** and check that coverage grows toward the distance and stays off the subject. - Select the original image and the Image Compositor together to compare them as A and B in the Design panel. ## Variations - **Video shots.** Use **Depth Anything V3 Video** to make a depth clip, and the **Compositor** (video) node instead of Image Compositor. - **Haze without refocus.** Skip ZDefocus and connect the image straight to Depth Atmosphere. Depth Atmosphere also works outside a compositor. - **Limit the effect to part of the frame.** Both nodes accept an optional mask; set **Mask Source** and **Invert Mask** as needed. - **Generated stills.** Start from an [Image Generator](https://docs.nodespell.com/creative-tools/image-generator/) result and keep its take fixed before you build the depth chain. ## Frequently asked questions ### Why is the wrong part of the image blurred? The depth encoding is probably inverted. Both Nodespell depth nodes put near surfaces in white, so ZDefocus and Depth Atmosphere should use **White Near**. ### Does changing ZDefocus or Depth Atmosphere use credits? No. They are image-processing nodes. Only generating the depth map uses credits. ### Why can't I add ZDefocus in the main graph? ZDefocus is available only inside a compositor. Enter an Image Compositor or Compositor first. Depth Atmosphere works in both places. ### Is the depth map a real distance? No. Both depth nodes produce relative depth, not distances in meters. Use **Depth Min** and **Depth Max** on ZDefocus or Depth Atmosphere to remap the range you need. ## Related - [Depth and defocus](https://docs.nodespell.com/edit/depth-and-defocus/) — every depth, defocus, and atmosphere control - [Compositing](https://docs.nodespell.com/edit/compositing/) — compositor inputs and outputs - [Viewers](https://docs.nodespell.com/edit/viewers/) — inspect channels and compare results - [Composite and finish existing images](https://docs.nodespell.com/recipes/composite-and-finish/) — the compositor basics --- # Composite and finish existing images > Key a foreground, color-correct and position it, merge it over a background in a Nodespell Image Compositor, then crop and download the finished frame. Source: https://docs.nodespell.com/recipes/composite-and-finish/ · Last updated: 2026-11-10 The Nodespell composite-and-finish recipe builds one finished still from images you already have. You bring a foreground and a background into an Image Compositor, pull a matte with Keyer, match the foreground with Color Correct, place it with Transform, merge it over the background, then crop and download. Compositor nodes run in the editor, so this recipe uses no generation credits. **What you need** - A foreground image (for example, a subject shot on a green or blue screen, or a cutout from [Remove Background](https://docs.nodespell.com/creative-tools/remove-background/)). - A background image. - Both images in [Asset Manager](https://docs.nodespell.com/assets/asset-manager/), or as results on nodes in your Project. **Graph at a glance:** Background Asset + Foreground Asset → Image Compositor { Compositor Input → Keyer → Color Correct → Transform → Merge → Compositor Output } → Crop → download > **Credits:** Keyer, Color Correct, Transform, Merge, and Crop are image-processing nodes, not model runs, so they do not use credits. You spend credits only if you generate a missing ingredient, such as a background from Image Generator. ## Stage 1: Bring the images into an Image Compositor 1. Open the Project in **Graph** mode. 2. Press `Tab` and add an **Asset** node for the background. Right-click it, choose **Load Asset**, and pick the background image. Repeat for the foreground. 3. Press `Tab` and add **Image Compositor**. 4. Connect the background Asset to the Image Compositor **Images** input first, then connect the foreground Asset. The connection order sets each image's index: the first is 0, the second is 1. 5. Select the Image Compositor node and set **Canvas size**, for example **First input** to match the background, or a preset such as **1080p Landscape · 1920 × 1080**. Image Compositor is the still-image compositor; **Compositor** is the time-based version for video. See [Compositing](https://docs.nodespell.com/edit/compositing/). ## Stage 2: Pull the foreground matte 1. Select the Image Compositor and press `Enter`, or right-click it and choose **Enter Node**. A new compositor already contains a **Compositor Input** connected to a **Compositor Output**. 2. Select the existing Compositor Input and leave **Index** at `0`. This is the background. 3. Press `Tab` and add a second **Compositor Input**. Set its **Index** to `1`. This is the foreground. 4. Press `Tab`, add **Keyer**, and connect the foreground Compositor Input to it. 5. Choose the keying mode that matches the footage: **Green Screen**, **Blue Screen**, **Red Screen**, **Chroma Distance**, **Luminance**, or **Saturation**. 6. Adjust **Screen Color**, **Clip Black**, and **Clip White** until the background is clean and the subject is solid. Set **Output** to **Matte View** while you tune, then back to **Write Alpha**. If the foreground is already a cutout with transparency, skip Keyer. Add **Erode/Dilate** after Keyer when the matte edge needs to shrink or grow. See [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/). ## Stage 3: Match the color and place the foreground 1. Add **Color Correct** after Keyer. Choose **Mode** (**Lift Gamma Gain** or **Shadows Midtones Highlights**) and use **Color Wheels** to match the foreground's blacks, midtones, and highlights to the background. 2. Add **Transform** after Color Correct. Scale, move, and rotate the foreground into place. Choose a **Filter** (for example **Lanczos 3**) for sharper resampling. See [Color correction](https://docs.nodespell.com/edit/color-correction/) and [Transforms and crop](https://docs.nodespell.com/edit/transforms-and-crop/). ## Stage 4: Merge and output 1. Add **Merge**. 2. Connect the background Compositor Input to **Background** and the Transform output to **Foregrounds**. 3. Leave **Mode** on **Over** to place the foreground on top. Lower the mix to blend it. 4. Connect the Merge output to **Compositor Output**, replacing the default connection from the first Compositor Input. 5. Press `U` to return to the parent graph. **Merge** offers many blend modes, including **Screen**, **Multiply**, **Overlay**, and a **Custom** mode with expressions for **Colour** and **Alpha**. See [Compositing](https://docs.nodespell.com/edit/compositing/). ## Stage 5: Crop and download 1. Back in the parent graph, press `Tab`, add **Crop**, and connect the Image Compositor **Image** output to it. 2. Select Crop. Drag the handles in the Design panel viewer to set the frame. Hold `Shift` to lock the aspect ratio and `Option/Alt` to resize from the center. 3. Right-click the Crop node and choose **Download**. See [Export](https://docs.nodespell.com/edit/export/) for download options. ## Check the result - Select the Image Compositor. The Design panel shows the composite, and you can zoom to inspect the matte edge. - Temporarily set Keyer **Output** to **Matte View** and look for gray holes in the subject or specks in the background. - Select the original foreground Asset and the Crop node together to compare them as A and B in the Design panel. ## Variations - **Several foregrounds.** Connect more images to **Images** (indices 2, 3, …) and feed each prepared layer into **Foregrounds** of the same Merge. - **Grade the whole frame.** Add Color Correct or **Grade** after Merge, inside the compositor, so the correction applies to the assembled image. - **Composite a video shot.** Use **Compositor** instead of Image Compositor when the background or foreground is a video, then render and download its video output. - **Add depth effects.** Feed a depth map into the same compositor and add ZDefocus or Depth Atmosphere; see [Refocus with depth and add atmosphere](https://docs.nodespell.com/recipes/depth-refocus-and-atmosphere/). ## Frequently asked questions ### Why does my compositor show only the background? The default Compositor Output is connected to the first Compositor Input. Connect the Merge output to **Compositor Output** instead. ### How does a Compositor Input know which image to use? **Index** is a zero-based position in the parent compositor's **Images** input, in connection order. The first connected image is `0`. ### Can I use Keyer outside a compositor? Yes. Keyer, Color Correct, Transform, and Crop also work in the main graph. Merge and Compositor Input work only inside a compositor. ### How do I get back out of the compositor? Press `U` to go up one level to the parent graph. ## Related - [Compositing](https://docs.nodespell.com/edit/compositing/) — compositor types, inputs, and Merge modes - [Masks and keying](https://docs.nodespell.com/edit/masks-and-keying/) — Keyer and matte tools in depth - [Color correction](https://docs.nodespell.com/edit/color-correction/) — Color Correct, Curves, and Hue Correct - [Transforms and crop](https://docs.nodespell.com/edit/transforms-and-crop/) — Transform, Resize, and Crop - [Export](https://docs.nodespell.com/edit/export/) — download finished media --- # Revise an image or shot without starting over > Edit, vary, rerun, and switch models in Nodespell while every earlier take stays available, so one change never costs you the work around it. Source: https://docs.nodespell.com/recipes/revise-without-starting-over/ · Last updated: 2026-11-10 Revising in Nodespell means changing one step of a graph while everything else stays as it was. Each node keeps its results as numbered takes, edits and variations become new nodes downstream of the original, and switching models creates a branch instead of overwriting. This recipe shows the four ways to revise: edit, vary, regenerate and choose takes, and branch to another model. **What you need** - A Project with at least one generated result you want to improve, such as an image from [Make your first image](https://docs.nodespell.com/start/first-image/). - A clear note of what should change and what must stay the same. - Credits for any new generation. Reusing existing takes is free. **Graph at a glance:** Original node → Image Editor or Variations (new node) · original node → more takes → **Use Take** · original node → model branch (new node) > **Where credits are spent:** Only generation uses credits: a new edit, a new variation, new takes, or a branch run. Choosing a take, keeping one, or running with existing takes does not. On paid plans, the Create composer and the run button show an estimate before you run. ## Stage 1: Make a targeted edit Use Image Editor when the image is nearly right and one thing must change. 1. In **Create**, choose **Image Editor**. 2. In **Source image**, select **Choose assets** and pick the result, then select **Use selection**. In Graph you can instead connect the original node's output to the Image Editor source input. 3. In **Prompt**, describe only the change and what must stay, for example "Replace the mug with a glass. Keep the lighting, framing, and table unchanged." 4. Select **Edit**. The edit is a new node. The original image and its takes are untouched, so you can compare before and after by selecting both nodes; the Design panel shows two selected images as A and B. See [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/). ## Stage 2: Try a variation Use Variations for a new angle or a directed change that keeps the subject consistent. 1. In **Create**, choose **Variations** and set **Source image** to the result. 2. Choose a **Variation mode**: - **Reframe** — choose a new framing or camera angle in **Shot type**, such as **Close-up**, **Low angle**, or **Over-the-shoulder**. - **Custom** — write **Variation instructions** that say what should change and what should stay consistent. 3. Set **Generations** to try several at once, then select **Generate**. See [Variations](https://docs.nodespell.com/creative-tools/variations/). ## Stage 3: Generate more takes and choose the best one When the settings are right but the result is not, ask the same node for more takes. Every run of a node adds a numbered take; earlier takes stay. 1. In **Graph**, select the node. 2. Open **Run options** (the arrow beside the run button) and choose a batch such as **Generate 3 takes**, or select the primary button, which reads **Generate** when it will make one new take. 3. When the takes finish, open **Takes** beside the run button, or right-click the node and choose **Takes…**. 4. Select a take tile to preview it. Previewing never changes the output. 5. Select **Use Take** with its number to make it the node's output. Select **Back to the take in use** to cancel a preview. The **More take actions** menu controls what happens on later runs: | Action | What it does | | ------------------- | ------------------------------------------------------------------------------------------------------- | | **Keep this take** | The node stays on this take. New takes are added to the history but do not replace it. | | **Follow latest** | The node always uses the newest successful take. | | **Pin to a branch** | Adds a node below the original that always outputs this take, so you can build a separate path from it. | | **Delete Take** | Removes the take from the history after checking what uses it. Deleting does not refund credits. | See [Takes](https://docs.nodespell.com/graph/takes/) for every take control. ### Rerun downstream work without regenerating To refresh downstream nodes using what you already have, open **Run options** and choose **Run with existing takes**. Nodespell shows "Uses the takes you already have. Nothing is charged." For finer control, choose **Plan this run…** and set each node to **Reuse** or **Generate**. See [Run nodes](https://docs.nodespell.com/graph/run-nodes/). ## Stage 4: Branch to a different model Switching the model of a node that has already run creates a branch, so the original node, its takes, and everything connected after it stay as they are. 1. In **Create**, select the node you want to try on another model. 2. Pick a different model in the model picker. 3. In **Create a model branch?**, review which settings and connections are **Preserved** and which are **Not copied**. 4. Select **Create branch draft**. Select **Keep current model** to cancel. 5. Adjust any settings the new model needs, then select **Generate**. The draft "will create a new downstream model node; the previous node remains unchanged." If some settings cannot carry over, Nodespell lists them, for example **Image size** or **Aspect ratio**. See [Branch and switch presentation](https://docs.nodespell.com/graph/branch-and-switch-presentation/). ## Check the result - Select the original node and the revision together to compare them as A and B in the Design panel. - Check the **Uses** label in the node's takes: **Latest** means it follows new takes, **Kept take** means it stays put. - Confirm downstream nodes show the output you expect before you run anything paid after them. ## Variations - **Save a checkpoint first.** Before a large restructure, save a Project Snapshot so you can return to the whole graph. See [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/). - **Revise video the same way.** Modify Video and Extend Video also create new nodes from a source clip; see [Extend a shot](https://docs.nodespell.com/recipes/extend-a-shot/). - **Compare several models at once.** Feed one prompt into several model nodes; see [Compare models side by side](https://docs.nodespell.com/recipes/compare-models-side-by-side/). ## Frequently asked questions ### Does generating a new take delete the old one? No. Every run adds a numbered take and the earlier takes stay in the node's history until you delete them. ### Why did my chosen take get replaced after the next run? The node was following the latest take. Open **More take actions** and choose **Keep this take** so new takes are added without replacing it. ### Does changing the model overwrite my node? Not for a node that has already run. Nodespell asks **Create a model branch?** and makes a new node, leaving the original and its takes unchanged. ### Is choosing a different take charged? No. Using, keeping, or previewing a take, and **Run with existing takes**, do not generate anything and are not charged. ### Can I get credits back by deleting a take? No. Nodespell states that deleting doesn't refund credits or free storage, and it can't be undone there. ## Related - [Takes](https://docs.nodespell.com/graph/takes/) — preview, use, keep, pin, and delete takes - [Run nodes](https://docs.nodespell.com/graph/run-nodes/) — what runs, what is reused, and run plans - [Branch and switch presentation](https://docs.nodespell.com/graph/branch-and-switch-presentation/) — model branches in detail - [Image Editor](https://docs.nodespell.com/creative-tools/image-editor/) — instruction-led edits - [Variations](https://docs.nodespell.com/creative-tools/variations/) — reframes and custom variations --- # Compare AI models side by side with one prompt > Feed one Nodespell Prompt node into several image model nodes, run them together, compare results in the Design panel, and keep the best model. Source: https://docs.nodespell.com/recipes/compare-models-side-by-side/ · Last updated: 2026-11-10 Comparing models in Nodespell means sending the same prompt to several model nodes at once and judging the outputs next to each other. One Prompt node feeds every model, so the only difference between results is the model. You run them together, compare pairs as A and B in the Design panel, and keep the winner connected to the rest of your graph. **What you need** - A prompt you want to test. - A shortlist of two to four models. Use the [model picker](https://docs.nodespell.com/creative-tools/choose-a-model/) to find candidates. - Credits for one run per model. **Graph at a glance:** Prompt → Model A, Model B, Model C (in parallel) → Design panel compare → winner → next step > **Where credits are spent:** Each model node is a separate paid run, so comparing four models costs four runs. On paid plans, hover the run button before you start: it shows the estimate for everything selected and which nodes will generate. ## Stage 1: Write the prompt once 1. Open the Project in **Graph** mode. 2. Press `Tab` and add a **Prompt** node. 3. Double-click it and write the prompt. See [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/). ## Stage 2: Add the model nodes and connect them 1. Press `Tab` and type a model name, for example `Nano Banana 2` or `GPT Image`, and add the model node. Repeat for each model on your shortlist. 2. Connect the Prompt output to the **Prompt** input of every model node. One output can feed any number of inputs. 3. Select each model node and match the settings that affect the look, such as aspect ratio or image size, so the comparison is fair. Leave model-specific options at their defaults unless you are testing them. You can also start from a Creative Tool, such as **Image Generator**, and switch its model; but a separate model node per candidate keeps every result on the canvas at once. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/) and [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). ## Stage 3: Run every model at once 1. Select all the model nodes. 2. Hover the run button to see what will happen: each model node is marked to generate, and the Prompt node needs no run. 3. Press `Cmd/Ctrl+Shift+R` or select the run button. Progress for each node appears in the [Tasks panel](https://docs.nodespell.com/graph/tasks/). To get more than one result from each model, open **Run options** and choose a batch such as **Generate 2 takes**. ## Stage 4: Compare in the Design panel 1. Select exactly two of the model nodes. The Design panel shows them as **A** and **B** in a comparison view. 2. Open the comparison mode menu and choose how to view them: **Horizontal swipe**, **Vertical swipe**, **Side by side**, **Stacked**, **Flip**, **Dissolve**, or **Difference**. 3. Press `X` to swap A and B. 4. Zoom in on fine detail such as text, hands, and edges. 5. Select a different pair to continue. The winner of each pair goes on to the next round. The comparison view handles two images at a time. For a quick overview of all results, look at the node previews on the canvas. See [Viewers](https://docs.nodespell.com/edit/viewers/). ## Stage 5: Keep the winner 1. Connect the winning node's output to the next step, such as Relight, Image Upscaler, or Image to Video. 2. If the winning node has several takes, open its **Takes**, select **Use Take** on the best one, and choose **Keep this take**. 3. Disconnect or delete the losing model nodes so they are not run again by accident. ## Check the result - Confirm every model received the same prompt: each model node's **Prompt** input should be connected to the one Prompt node. - Compare at the same zoom level and in the same comparison mode for every pair. - Note the run estimate per model if cost matters in your decision. ## Variations - **Compare video models.** Connect the Prompt and a start-frame image to several image-to-video model nodes. The Design panel compare works for images; open each video take with **View Full Screen** to judge motion. - **Compare prompts on one model.** Do the reverse: several Prompt nodes, each feeding a copy of the same model node. - **Compare in Create.** In Create, use **Previous model** and **Next model** in the model picker to try models one after another on one node. Each change on a node that has already run creates a branch, so earlier results stay. See [Revise without starting over](https://docs.nodespell.com/recipes/revise-without-starting-over/). - **Add references to the test.** Connect the same reference image to every model that accepts one, to compare how well each keeps a subject; see [Keep a character or product consistent](https://docs.nodespell.com/recipes/consistent-character/). ## Frequently asked questions ### How many models can I compare at once? You can run as many model nodes as you select. The Design panel compares two selected images at a time, so judge them in pairs. ### Why does one model ignore my prompt? Check that its **Prompt** input is connected to the Prompt node. Some models also have required inputs, such as a source image, and cannot run without them. ### Does rerunning the comparison regenerate every model? Running a selection makes new takes for the selected model nodes. To refresh only one model, select only that node before you run. ### Can I compare two takes from the same model? Yes. Open the node's **Takes** to preview each take in turn, or use **Pin to a branch** on one take so it appears as its own node, then select both nodes to compare them. ## Related - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/) — find and switch models - [Viewers](https://docs.nodespell.com/edit/viewers/) — Design panel comparison modes and zoom - [Run nodes](https://docs.nodespell.com/graph/run-nodes/) — run previews, estimates, and batches - [Takes](https://docs.nodespell.com/graph/takes/) — keep the winning result --- # Nodespell reference > Lookup tables for Nodespell. Panels, keyboard shortcuts, supported media, glossary, troubleshooting, frequently asked questions and release notes. Source: https://docs.nodespell.com/reference/ · Last updated: 2026-11-10 The Nodespell reference is the lookup half of the documentation: complete tables rather than walkthroughs. Use it when you already know what you are doing and need an exact name, input, shortcut, format or definition. For step-by-step learning, start with [Make your first image](https://docs.nodespell.com/start/first-image/) or the [recipes](https://docs.nodespell.com/recipes/). ## The workspace - [Panels](https://docs.nodespell.com/reference/panels/): What each panel is for and which guide covers it. - [Keyboard shortcuts](https://docs.nodespell.com/reference/shortcuts/): Shortcuts for macOS and Windows, grouped by area. - [Supported media](https://docs.nodespell.com/reference/supported-media/): File types and size limits for uploads, and the formats Nodespell produces. ## Answers - [Glossary](https://docs.nodespell.com/reference/glossary/): One or two sentence definitions of every Nodespell term. - [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/): Symptom, likely cause and fix, grouped by area. - [Frequently asked questions](https://docs.nodespell.com/reference/faq/): Short answers to common questions about Nodespell. - [What's new](https://docs.nodespell.com/whats-new/): Dated release notes: new features, improvements and fixes. ## For AI assistants Every page on this site is also published as Markdown. Replace the trailing slash of a page address with `.md`, for example `/graph/takes.md`. Two index files describe the whole site: | File | Contents | | ---------------------------------- | -------------------------------------------------------------- | | [`/llms.txt`](https://docs.nodespell.com/llms.txt) | A structured index of every guide with a one-line description. | | [`/llms-full.txt`](https://docs.nodespell.com/llms-full.txt) | The full text of every guide in one file, in navigation order. | ## Where is the list of models and nodes? The full catalog of AI models and nodes lives inside Nodespell, not on this site. Sign in and open the **Library** panel to browse every node with its inputs, outputs and parameters, or open the model picker in any Creative Tool to compare models. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/) and [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). --- # Nodespell panels reference > Look up every panel in the Nodespell workspace, from Graph and Parameters to Assets, Design, Library, Tasks and Assistant, and how to open each one. Source: https://docs.nodespell.com/reference/panels/ · Last updated: 2026-11-10 Panels are the tools that make up the Nodespell workspace: **Graph** for the node canvas, **Parameters** for settings, **Assets** for media, **Design** for viewing and editing a node's output, and more. Each panel is listed with what it is for and the guide that covers it in depth. ## How do you open a panel? In **Graph** mode: 1. Select **+** at the end of any pane's tab bar, or **+** inside an edge dock. 2. Choose the panel from the menu. You can also open a panel as a floating window from **Window** > **New Panel Window**, or press `Cmd/Ctrl+K` and search for **Add** followed by the panel name, such as **Add Tasks Panel**. In **Immersive**, the right tool rail opens the most used panels as quick panels. In **Create**, the composer and the **Assistant** and **Assets** rail items take their place. See [Tour the interface](https://docs.nodespell.com/start/interface/) for panes, tabs and docks. ## All panels The **+** menu groups panels the same way as this table. | Group | Panel | Use it to | Guide | | ---------- | -------------- | -------------------------------------------------------------------------------------------------- | ------------------------------------------------ | | Essential | **Graph** | Add, arrange, connect and run nodes on the canvas. | [Graph Editor](https://docs.nodespell.com/graph/graph-editor/) | | Essential | **Parameters** | Edit the settings of the selected node. | [Parameters](https://docs.nodespell.com/graph/parameters/) | | Essential | **Assets** | Browse, upload, search, tag, preview and organize media, and drag it into the graph. | [Asset Manager](https://docs.nodespell.com/assets/asset-manager/) | | Essential | **Design** | View and edit the selected node's output: viewers, compositor, crop and Timeline editors. | [Viewers](https://docs.nodespell.com/edit/viewers/) | | Library | **Library** | Find nodes, Workflows, shared Projects and collections in one place. | [Add nodes](https://docs.nodespell.com/graph/add-nodes/) | | Library | **My Nodes** | List the custom nodes in your account. | [Add nodes](https://docs.nodespell.com/graph/add-nodes/) | | Automation | **Tasks** | Check queued, running, completed and failed tasks, their timing and credit changes. | [Tasks](https://docs.nodespell.com/graph/tasks/) | | Automation | **Assistant** | Talk to the Agent about the current graph and ask it to make changes. | [Agent](https://docs.nodespell.com/agent/) | | Advanced | **References** | See which nodes use the selected node, through reference nodes or expressions. | [Expressions](https://docs.nodespell.com/graph/expressions/) | | Advanced | **Tree View** | Browse every node in the Project as a tree of Levels and Folders, filter by name and select nodes. | [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/) | | Advanced | **Console** | Read log messages, warnings and errors, filter them and copy them for support. | [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/) | ## Graph The **Graph** panel is the node canvas where you build the Project. It shows one Level of the graph at a time, with the run controls floating near the top. You can open several **Graph** panels, for example to watch two Levels side by side; give their tabs different colors to tell them apart. See [Graph Editor](https://docs.nodespell.com/graph/graph-editor/). ## Parameters The **Parameters** panel shows the parameters of the selected node. A node presented as a Creative Tool shows the same guided controls as the Create composer, including its model picker. See [Parameters](https://docs.nodespell.com/graph/parameters/). ## Assets The **Assets** panel is the Asset Manager. It holds everything you upload and the results you save, in **Personal Assets** or **Team Assets**. Drag an Asset onto the graph to create an Asset node, or select an Asset and press `Space` to preview it. You can also open it from **Account** > **Assets**. See [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ## Design The **Design** panel shows a larger, interactive view of the selected node. For an image, video or text output it shows a viewer where you can zoom, inspect color channels, compare two selected images and change the preview resolution; text outputs offer search, wrapping and a Markdown view. Compositor nodes show their compositor canvas, and the **Timeline** node shows the Timeline editor. If no node is selected, the panel asks you to select one. See [Viewers](https://docs.nodespell.com/edit/viewers/), [Compositing](https://docs.nodespell.com/edit/compositing/) and [Timeline](https://docs.nodespell.com/edit/timeline/). ## Library The **Library** panel collects everything reusable. Choose a scope, **All**, **Nodes**, **Workflows**, **Projects** or **Collections**, and a source such as **Official**, **Mine**, **This project**, **Team**, **Public** or **Favourites**. Search, filter, sort and switch between **Grid** and **List** layouts, then open an item to see its details or place it in your graph. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/) and [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/). ## Tasks The **Tasks** panel lists the AI tasks your runs submitted. Filter by status (**Completed**, **Running**, **Queued**, **Failed**, **Pending**) or by model type, search, and select **Show Details** for start time, end time, duration, error and credit change. Other tabs show credit usage over time. See [Tasks](https://docs.nodespell.com/graph/tasks/). ## Assistant The **Assistant** panel is where you work with the Nodespell [Agent](https://docs.nodespell.com/agent/). Describe what you want, choose a mode (**Ask**, **Plan** or **Edit**), and review what it proposes before it changes your graph. The same panel opens from the **Assistant** item in the Create rail and the Immersive right rail. ## References, Tree View and Console These are inspection panels for larger Projects. - **References** lists the nodes that use the selected node, grouped into **Reference nodes** and **Expressions**, with **Show in graph** for each. - **Tree View** lists every node in the Project in a tree, with **Expand all**, **Collapse all**, a name filter and **Select in project**. - **Console** shows log messages with level filters (**All**, **Errors**, **Warnings**, **Info**), a search field and **Copy visible messages**. The status bar's console summary opens it as a window. ## Panels in Create and Immersive | Need | Create | Immersive | | ------------ | -------------------------------------------- | -------------------------------------------------------- | | Graph | Shown on the right of the composer | Fills the window | | Settings | Composer controls | **Parameters** on the right rail when a node is selected | | Media | **Assets** in the Create tools rail | **Assets** on the right rail | | Output view | Results in the composer, **View fullscreen** | **Design** on the right rail when a node is selected | | Agent | **Assistant** in the Create tools rail | **Assistant** on the right rail | | Logs | Not available | **Console** on the right rail | | Other panels | Not available | **Window** > **New Panel Window** | ## Frequently asked questions ### Where is the Node Browser? Node search and details are in the **Library** panel under the **Nodes** scope. The command palette's **Show Node in Browser** opens it on the selected node, and the `Tab` menu searches nodes directly on the canvas. ### Why is the Design panel empty? The **Design** panel follows the node selection. Select a node in the **Graph** panel to show its output. Comparing two images needs two image nodes selected. ### Can I open two copies of the same panel? Yes. Right-click a panel's tab and choose **Duplicate**, or add the panel again from **+**. This is useful for two **Graph** panels showing different Levels. ### Why is the Assistant panel called Assistant and not Agent? **Assistant** is the panel label in the product. The docs call the feature the Agent; both names refer to the same in-product assistant. ## Related - [Tour the interface](https://docs.nodespell.com/start/interface/) — panes, tabs, docks and layouts. - [Choose a workspace mode](https://docs.nodespell.com/start/workspace-modes/) — how panels differ by mode. - [Graph Editor](https://docs.nodespell.com/graph/graph-editor/) — the main canvas. - [Asset Manager](https://docs.nodespell.com/assets/asset-manager/) — media in depth. - [Shortcuts](https://docs.nodespell.com/reference/shortcuts/) — keyboard shortcuts for every panel. --- # Keyboard shortcuts > Every Nodespell keyboard shortcut for macOS and Windows or Linux, grouped by area: app, graph navigation, nodes, running, viewers, Assets and Timeline. Source: https://docs.nodespell.com/reference/shortcuts/ · Last updated: 2026-11-10 Nodespell keyboard shortcuts let you navigate the graph, create and arrange nodes, run work and control viewers without the mouse. On macOS the main modifier is `Cmd`; on Windows and Linux it is `Ctrl`. `Option` on macOS is `Alt` on Windows and Linux. Graph shortcuts act on the graph panel you last clicked, and no shortcut fires while you are typing in a text field. To see the shortcuts in the app, open **Preferences** › **Hotkeys**, or turn on **Status Bar Hotkey Help** in View Settings for hints in the status bar. ## App-wide | Action | macOS | Windows and Linux | | ------------------------- | ------------- | ----------------- | | Open the command palette | `Cmd+K` | `Ctrl+K` | | Open Preferences | `Cmd+,` | `Ctrl+,` | | Find in graph | `Cmd+F` | `Ctrl+F` | | Find and replace in graph | `Cmd+Shift+F` | `Ctrl+Shift+F` | | Undo | `Cmd+Z` | `Ctrl+Z` | | Redo | `Cmd+Shift+Z` | `Ctrl+Shift+Z` | | Toggle top dock | `Cmd+1` | `Ctrl+1` | | Toggle bottom dock | `Cmd+2` | `Ctrl+2` | | Toggle left dock | `Cmd+3` | `Ctrl+3` | | Toggle right dock | `Cmd+4` | `Ctrl+4` | | Close open docks | `Esc` | `Esc` | ## Graph: move around | Action | macOS | Windows and Linux | | -------------------------------------------------------- | ------------------------------ | -------------------------------- | | Pan | `←` `→` `↑` `↓` | `←` `→` `↑` `↓` | | Pan faster | `Shift` + arrow | `Shift` + arrow | | Zoom in / out | `=` / `-` | `=` / `-` | | Frame selected nodes | `F` | `F` | | Frame all nodes | `A` | `A` | | Overview: hold to see the whole graph, release to return | Hold `Z` | Hold `Z` | | Reset zoom | `0` | `0` | | Reset pan | `Shift+0` | `Shift+0` | | Previous / next view | `Shift+[` / `Shift+]` | `Shift+[` / `Shift+]` | | Go inside the selected Level, Folder or container | `Enter` | `Enter` | | Go up one level | `U` | `U` | | Jump to a quick-jump slot | `Cmd+Shift+1` to `Cmd+Shift+0` | `Ctrl+Shift+1` to `Ctrl+Shift+0` | | Save the selection to a quick-jump slot | `Option+1` to `Option+0` | `Alt+1` to `Alt+0` | | Show quick-jump labels | Hold `Cmd` | Hold `Ctrl` | On a trackpad, pan with a two-finger drag and zoom with a pinch. See [Graph Editor](https://docs.nodespell.com/graph/graph-editor/) for mouse and trackpad gestures. ## Graph: select | Action | macOS | Windows and Linux | | -------------------------------------- | ----------------------- | ----------------------- | | Select all | `Cmd+A` | `Ctrl+A` | | Clear selection | `Esc` | `Esc` | | Add to or remove from the selection | `Shift` + click or drag | `Shift` + click or drag | | Select upstream nodes | `Shift+U` | `Shift+U` | | Select downstream nodes | `Shift+D` | `Shift+D` | | Select connected nodes | `Shift+C` | `Shift+C` | | Move the selection along connections | `Cmd` + arrow | `Ctrl` + arrow | | Extend the selection along connections | `Cmd+Shift` + arrow | `Ctrl+Shift` + arrow | | Previous / next selection in history | `[` / `]` | `[` / `]` | | Switch selection tool | `Shift+B` | `Shift+B` | ## Graph: create and edit nodes | Action | macOS | Windows and Linux | | -------------------------------------------------------------- | --------------------------- | ------------------------------ | | Open the Tab menu at the pointer | `Tab` | `Tab` | | Add a Prompt node at the pointer | `P` | `P` | | Add an Asset at the pointer | `O` | `O` | | Add a Dot (connected to the selected node, if one is selected) | `.` | `.` | | Create a Group | `Cmd+G` | `Ctrl+G` | | Copy / cut / paste | `Cmd+C` / `Cmd+X` / `Cmd+V` | `Ctrl+C` / `Ctrl+X` / `Ctrl+V` | | Delete selected nodes | `Delete` or `Backspace` | `Delete` or `Backspace` | | Duplicate while dragging | Hold `Option` and drag | Hold `Alt` and drag | | Drag a node with its upstream nodes | Hold `Cmd` and drag | Hold `Ctrl` and drag | | Bypass the selected node | `B` | `B` | | Show node info | `I` | `I` | | Open the selected node's viewer | `Space` | `Space` | | Disconnect selected nodes | `X` | `X` | | Swap input connections | `Shift+R` | `Shift+R` | | Cut connections: hold and drag across them | Hold `Y` | Hold `Y` | | Show dots on connections | Hold `Option` | Hold `Alt` | | Cancel a connection in progress | `Esc` | `Esc` | | Cycle the connection line style | `Shift+S` | `Shift+S` | | Toggle the docked Parameters panel | `Q` | `Q` | | Toggle the model hub | `Shift+M` | `Shift+M` | ## Graph: arrange | Action | macOS | Windows and Linux | | ------------------------------------ | ----------------------- | --------------------- | | Tidy up selected nodes | `L` | `L` | | Adjust layout (layout mode) | `Shift+L` or `Option+L` | `Shift+L` or `Alt+L` | | Align vertically / horizontally | `V` / `H` | `V` / `H` | | Distribute vertically / horizontally | `Shift+V` / `Shift+H` | `Shift+V` / `Shift+H` | | Nudge the selection | `Option` + arrow | `Alt` + arrow | | Nudge the selection further | `Option+Shift` + arrow | `Alt+Shift` + arrow | | Make selected nodes larger / smaller | `Shift+=` / `Shift+-` | `Shift+=` / `Shift+-` | ## Running | Action | macOS | Windows and Linux | | ------------------------------------------------- | -------------- | ----------------- | | Run selected nodes (**Run Selected Tasks**) | `Cmd+Shift+R` | `Ctrl+Shift+R` | | Discard selected results (projects without takes) | `Cmd+Shift+C` | `Ctrl+Shift+C` | | Toggle result lock (projects without takes) | `Ctrl+Shift+L` | `Ctrl+Shift+L` | See [Run nodes](https://docs.nodespell.com/graph/run-nodes/) for what a run does. ## Immersive | Action | macOS | Windows and Linux | | ------------------------------ | ----- | ----------------- | | Show or hide Parameters | `Q` | `Q` | | Show or hide the Asset Manager | `T` | `T` | | Show or hide the Design panel | `W` | `W` | ## Parameters | Action | macOS | Windows and Linux | | ----------------------------------------------------- | ----- | ----------------- | | Set or remove a keyframe on the focused parameter row | `K` | `K` | ## Fullscreen image viewer | Action | Key | | -------------------------------- | ---------- | | Previous / next image | `←` / `→` | | Previous / next take (or result) | `↑` / `↓` | | Jump to image 1–9 | `1` to `9` | | Zoom in / out | `=` / `-` | | Reset view | `0` | | Home (reset zoom and pan) | `H` | | Toggle fullscreen | `F` | | Grid view | `G` | | Compare two images | `C` | | Toggle info | `I` | | Open in a new tab | `O` | | Close | `Esc` | ## Fullscreen video viewer | Action | Key | | -------------------------------------------- | --------- | | Play / pause | `Space` | | Step one frame back / forward (single video) | `←` / `→` | | Previous / next take (or result) | `↑` / `↓` | | Set loop in / out point | `[` / `]` | | Mute | `M` | | Zoom in / out | `=` / `-` | | Reset view / Home | `0` / `H` | | Toggle fullscreen | `F` | | Grid view | `G` | | Toggle info | `I` | | Open in a new tab | `O` | | Close | `Esc` | ## Fullscreen audio, text and 3D viewers | Viewer | Action | Key | | --------------- | ------------------------------------ | --------------------- | | Audio | Play / pause | `Space` | | Audio | Seek back / forward 5 seconds | `Shift+←` / `Shift+→` | | Audio | Toggle loop | `L` | | Audio | Toggle normalize | `N` | | Text | Switch between Markdown and raw view | `M` | | Text | All versions / back to current | `G` | | Text | Compare | `C` | | Text | Copy all text | `Shift+C` | | Text | Larger / smaller text | `=` / `-` | | Text | Toggle info / open in a new tab | `I` / `O` | | Audio, text, 3D | Previous / next take (or result) | `↑` / `↓` | | All | Close | `Esc` | ## Asset Manager | Action | macOS | Windows and Linux | | ----------------------- | ------- | ----------------- | | Preview selected Assets | `Space` | `Space` | | Copy selected Assets | `Cmd+C` | `Ctrl+C` | | Clear selection | `Esc` | `Esc` | ## Timeline These work while the Timeline has focus. Press `?` in the Timeline to see them in the app. | Action | macOS | Windows and Linux | | ---------------------------------------------------------------------- | --------------------------- | ------------------------------ | | Play / pause | `Space` | `Space` | | Reverse / stop / forward (hold `K` and tap `J` or `L` to step a frame) | `J` / `K` / `L` | `J` / `K` / `L` | | Select / Blade / Trim tool | `A` / `B` / `T` | `A` / `B` / `T` | | Split at playhead | `S` | `S` | | Delete / ripple delete | `Delete` / `Shift+Delete` | `Delete` / `Shift+Delete` | | Duplicate | `Cmd+D` | `Ctrl+D` | | Copy / cut / paste | `Cmd+C` / `Cmd+X` / `Cmd+V` | `Ctrl+C` / `Ctrl+X` / `Ctrl+V` | | Select all | `Cmd+A` | `Ctrl+A` | | Disable clip | `D` | `D` | | Back / forward one frame | `←` / `→` | `←` / `→` | | Back / forward one second | `Shift+←` / `Shift+→` | `Shift+←` / `Shift+→` | | Previous / next cut | `↑` / `↓` | `↑` / `↓` | | Go to start / end | `Home` / `End` | `Home` / `End` | | Set range start / end | `I` / `O` | `I` / `O` | | Clear range | `Option+X` | `Alt+X` | | Fit timeline to selection or range | `F` | `F` | | Fit track height | `Shift+F` | `Shift+F` | | Edit properties | `Enter` | `Enter` | | Show keyboard shortcuts | `?` | `?` | See [Timeline](https://docs.nodespell.com/edit/timeline/) for editing in depth. ## Frequently asked questions ### Why does a graph shortcut do nothing? Click the graph panel first so it has focus, and make sure you are not typing in a text field or have a menu or dialog open. Graph shortcuts also pause while a fullscreen viewer is open. ### Can I see the shortcuts inside Nodespell? Yes. Open **Preferences** with `Cmd+,` or `Ctrl+,` and choose **Hotkeys** to search the full list. The Timeline shows its own list when you press `?`. ### What is the shortcut to run nodes? `Cmd+Shift+R` on macOS and `Ctrl+Shift+R` on Windows and Linux run the selected nodes. ### How do I pan without a middle mouse button? Use the arrow keys, a two-finger drag on a trackpad, or hold `Z` to see the whole graph and release to return. ## Related - [Graph Editor](https://docs.nodespell.com/graph/graph-editor/): navigation with mouse and trackpad - [Run nodes](https://docs.nodespell.com/graph/run-nodes/): the run shortcut in context - [Viewers](https://docs.nodespell.com/edit/viewers/): fullscreen viewers and the Design panel - [Timeline](https://docs.nodespell.com/edit/timeline/): editing clips with the keyboard - [Interface tour](https://docs.nodespell.com/start/interface/): command palette and docks --- # Supported file types, sizes and output formats > Reference tables for the file types Nodespell accepts for upload, the per-file size limits, and the formats Nodespell produces when you render or export. Source: https://docs.nodespell.com/reference/supported-media/ · Last updated: 2026-11-10 Nodespell accepts images, video, audio, documents, archives and 3D files as uploads, up to 200 MB per file and 50 files per upload. Uploaded files become Assets that you can preview, connect to nodes, and download. This reference lists the accepted types, the limits, the formats Nodespell outputs, and what is not supported. ## Upload limits | Limit | Value | | ------------------------ | -------------------------------------------------------------------------------------- | | Maximum size per file | 200 MB | | Maximum files per upload | 50 | | Storage | The workspace allowance; see [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/) | Individual models can have tighter input limits, such as a maximum clip length or file size. The model's own fields and descriptions state them. ## File types accepted by the Upload button The Asset Manager **Upload** button and the **Upload** choice in **Add asset** open a file picker that lists these types: | Kind | Formats | | ------------------ | ---------------------------------------------------------------------------------------------------------------------------------------- | | Image | JPEG, PNG, GIF, WebP, SVG, BMP, ICO, AVIF, HEIC, HEIF, TIFF | | Video | MP4, MOV (QuickTime), WebM, OGG video, AVI, MPEG, MKV | | Audio | MP3, WAV, OGG, WebM audio, AAC, AIFF | | Text and documents | TXT, HTML, CSS, JavaScript, JSON, XML, PDF, DOC, DOCX, XLS, XLSX, PPT, PPTX, Nodespell camera solve (`.nscam`) | | Archives | ZIP, RAR, 7Z, TAR, GZIP, BZIP2 | | 3D | OBJ, FBX, glTF, GLB, DAE, 3DS, BLEND, PLY, Alembic (ABC), STL, USD, USDA, USDC, USDZ, X3D, Gaussian splat (SPLAT, KSPLAT, SPZ, SOG), RAD | ## Extra types accepted by dropping on the graph Dropping files onto the Graph canvas also accepts other formats Nodespell recognizes by extension, including: | Kind | Additional formats | | ----- | --------------------------------------------------------------------- | | Image | EXR | | Video | MPG, FLV, WMV, 3GP, M4V | | Audio | FLAC, M4A, WMA, AU, WEBA | | Text | Markdown, CSV, YAML, LOG, INI, CFG, CHAN, CUBE, and common code files | Files that Nodespell does not recognize are refused with **Unsupported file types**. ## Which files preview in the browser? | File | Preview | | ----------------------------------------------- | ---------------------------------------------------------------------- | | JPEG, PNG, GIF, WebP, SVG, BMP, ICO, AVIF, TIFF | Shown in the browser | | HEIC, HEIF | Stored and usable; your browser may not display them | | EXR | Stored and usable; the Asset Manager shows **EXR preview unavailable** | | Video and audio | Play in the browser when your browser supports the codec | | Documents and archives | No preview; open or download the file | | 3D | Shown in the 3D viewer for supported formats | ## Formats Nodespell outputs | Source | Output format | | ---------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------- | | Downloads from nodes, viewers and the Asset Manager | The stored file, unchanged, at full resolution | | [Timeline](https://docs.nodespell.com/edit/timeline/) render | WEBM (default) or MP4 when your browser can encode H.264 and AAC | | **Trim Video**, **Join Video**, **Resize Video**, **Retime Video**, **Reverse Video**, **Transform Video** | MP4 | | **Transcode Video** | MP4, MOV, WebM or MKV | | **Extract Frame** | An image | | Compositor **Result** | Review PNG, review video, master linear EXR, or HDR review 10-bit WebM PQ where supported | | Compositor **Export EXR sequence…** | A folder of EXR frames (desktop Chrome or Edge) | | 3D Asset export | Alembic (.abc) | | AI model nodes | The format each model returns; shown in the model's description in the app | ## Video processing limits | Feature | Limit | | ---------------------------- | ------------------------------------------------- | | Timeline sequence size | 16 to 16,384 px per side | | Timeline frame rate | 1 to 240 fps (render frame rate 1 to 120 fps) | | Timeline render bitrate | 500 to 50,000 kb/s video; 16 to 512 kb/s audio | | **Transcode Video** | SDR video up to 256 MiB; transparency is not kept | | **Resize Video** custom size | 2 to 8,192 px per side | ## What is not supported? - Files larger than 200 MB through upload. - Pasting files from the clipboard into the Asset Manager or the graph. (The [Agent](https://docs.nodespell.com/agent/context-and-references/) accepts pasted attachments.) - Dragging files from your computer straight into the Asset Manager panel; use **Upload** or drop on the graph. - Bundled ZIP downloads of several Assets. - Keeping transparency through **Transcode Video**. - The Timeline editor in browsers without WebCodecs and the related storage features; use an up-to-date Chromium- or Firefox-based desktop browser. ## Frequently asked questions ### What is the maximum upload size in Nodespell? 200 MB per file, and up to 50 files in one upload. ### Can I upload EXR files? Yes, by dropping them onto the Graph canvas. They are stored and usable in nodes, but the Asset Manager does not show an EXR preview. ### Does Nodespell convert my uploads to another format? No. Nodespell stores the file you upload and downloads it unchanged. Thumbnails and previews are separate. ### Why won't my video play in the viewer? Playback uses your browser's decoders. If your browser cannot decode the video's codec, convert the file to a widely supported format, such as H.264 MP4, and upload it again. ## Related - [Upload and organize Assets](https://docs.nodespell.com/assets/upload-and-organize/): upload paths and duplicates - [Export finished media](https://docs.nodespell.com/edit/export/): download and render settings - [Video editing](https://docs.nodespell.com/edit/video-editing/): video nodes and their options - [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/): storage allowances - [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/): fixes for common problems --- # Nodespell glossary > Definitions of Nodespell terms, from Project, node, take and Level to Creative Tool, credits, edge dock and Agent, each linked to its full guide. Source: https://docs.nodespell.com/reference/glossary/ · Last updated: 2026-11-10 This glossary defines the terms used in Nodespell and in these docs. Each definition stands on its own and links to the guide that explains the feature in full. Labels shown in **bold** match the product interface. ## Workspace and modes ### Project A Project is the saved Nodespell workspace that holds a node graph, the takes its nodes have made, and the panel layout. See [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/). ### Project Snapshot A Project Snapshot is a named, saved state of a Project that you can open and restore later, created with **File** > **Save Version ...**. See [Projects and snapshots](https://docs.nodespell.com/projects/projects-and-snapshots/). ### Dashboard The dashboard is the Nodespell home screen where you find, create and open Projects in your personal or Team workspace. See [Dashboard](https://docs.nodespell.com/projects/dashboard/). ### Workspace mode A workspace mode is one of three ways to present a Project in Nodespell: **Create**, **Graph** or **Immersive**, chosen from the **Workspace mode** selector in the top bar. See [Workspace modes](https://docs.nodespell.com/start/workspace-modes/). ### Create Create is the Nodespell workspace mode for guided generation, where you pick a Creative Tool, choose a model, set its controls and select an action such as **Generate**. See [Workspace modes](https://docs.nodespell.com/start/workspace-modes/). ### Graph Graph is the Nodespell workspace mode for direct node editing, with the full system of panes, tabs, edge docks and saved layouts. See [Workspace modes](https://docs.nodespell.com/start/workspace-modes/). ### Immersive Immersive is the Nodespell workspace mode that gives one graph editor the whole window, with floating tool rails on the left and right and a collapsed top bar. See [Workspace modes](https://docs.nodespell.com/start/workspace-modes/). ### Composer The composer is the column in Create that shows the selected Creative Tool's model card, controls, **Generations** stepper and action button. See [Make your first image](https://docs.nodespell.com/start/first-image/). ## Interface ### Panel A panel is one tool in the Nodespell workspace, such as **Graph**, **Parameters**, **Assets** or **Design**, shown as a tab in a pane, a dock or a floating window. See [Panels reference](https://docs.nodespell.com/reference/panels/). ### Pane A pane is a rectangular area of the Graph-mode workspace that holds one or more panels as tabs; panes can be split, resized and maximized. See [Tour the interface](https://docs.nodespell.com/start/interface/). ### Edge dock An edge dock is a panel area that slides in from the top, bottom, left or right edge of the window in Graph mode, toggled with `Cmd/Ctrl+1` to `Cmd/Ctrl+4`. See [Tour the interface](https://docs.nodespell.com/start/interface/). ### Floating window A floating window is a panel moved out of the pane layout so it can sit anywhere over the workspace, created with **Tear off**, **Window** > **New Panel Window**, or by holding `Shift` while dragging a tab. See [Tour the interface](https://docs.nodespell.com/start/interface/). ### Layout A layout is a saved arrangement of panes, tabs, docks and tab colors that you can restore from **View** > **Layouts** or **Built-in Layouts** in Graph mode. See [Tour the interface](https://docs.nodespell.com/start/interface/). ### Panel colour A panel colour is a tint you give a panel's tab with **Colour** in the tab's right-click menu, saved with the layout. See [Tour the interface](https://docs.nodespell.com/start/interface/). ### Command palette The command palette is a searchable list of Nodespell commands, settings, nodes and tools, opened with `Cmd/Ctrl+K`. See [Tour the interface](https://docs.nodespell.com/start/interface/). ### Status bar The status bar is the strip along the bottom of the workspace that shows floating windows, shortcut **Hints** and the latest console message. See [Tour the interface](https://docs.nodespell.com/start/interface/). ### Design panel The Design panel shows a larger, interactive view of the selected node's output, including image and video viewers, the compositor canvas, crop handles and the Timeline editor. See [Viewers](https://docs.nodespell.com/edit/viewers/). ### Parameters panel The Parameters panel shows and edits the parameters of the selected node; it is also called the Parameter Editor. See [Parameters](https://docs.nodespell.com/graph/parameters/). ### Library The Library is the Nodespell panel that collects nodes, Workflows, shared Projects and collections, with search, sources and filters. See [Panels reference](https://docs.nodespell.com/reference/panels/). ### What's new What's new is the Nodespell release notes view, opened from **Help** > **What's new**, that lists new features, improvements and fixes by date. See [What's new](https://docs.nodespell.com/whats-new/). ## Graph building ### Node A node is one step in a Nodespell graph, such as an AI model, an edit, an effect, a media input or an organizer, with inputs, outputs and parameters. See [Core concepts](https://docs.nodespell.com/start/core-concepts/). ### Socket A socket is a connection point on a node: an output socket sends data and an input socket receives it. See [Connections](https://docs.nodespell.com/graph/connections/). ### Connection A connection is a directional link from one node's output socket to a compatible input socket on another node; Nodespell accepts it only when the types fit. See [Connections](https://docs.nodespell.com/graph/connections/). ### Parameter A parameter is a setting on a node, such as a prompt, a number, a choice, a color or a media input, edited in the Parameters panel or the Create composer. See [Parameters](https://docs.nodespell.com/graph/parameters/). ### Expression An expression is a formula typed into a parameter instead of a fixed value, so the value is calculated from other values in the graph. See [Expressions](https://docs.nodespell.com/graph/expressions/). ### Tab menu The Tab menu is the node search that opens at the pointer when you press `Tab` over the canvas; it finds nodes, models and Creative Tools and can connect the new node to the selected one. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/). ### Quick Add Quick Add is the section of the canvas right-click menu that lists commonly used nodes and Creative Tools by medium. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/). ### Node Browser The Node Browser is the searchable list of every node with its details, inputs, outputs and parameters, shown in the Library panel under **Nodes**. See [Add nodes](https://docs.nodespell.com/graph/add-nodes/). ### Prompt node A Prompt node holds reusable text that can feed the prompt of one or more model nodes; press `P` on the canvas to create one. See [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/). ### Saved text Saved text is your personal library of reusable prompts and text, inserted by typing `/` in a text field and managed in Preferences. See [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/). ### Asset node An Asset node places an image, video, audio clip or other Asset into the graph so other nodes can use it. See [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ### Group A Group is a frame around related nodes that organizes the graph without changing how it runs; Groups can be nested inside other Groups. See [Organize a graph](https://docs.nodespell.com/graph/organize/). ### Level A Level is a node that contains its own graph; press `Enter` to go inside and `U` to go back up. See [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/). ### Folder A Folder is a Level used only to organize nodes; it cannot run directly, and **Convert to Level** turns it back into a Level. See [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/). ### Flow direction Flow direction is the direction a graph is laid out in, top to bottom, bottom to top, left to right or right to left, set per Project or Level from the **Layout** menu. See [Organize a graph](https://docs.nodespell.com/graph/organize/). ### Tidy up Tidy up is the command that arranges nodes into clear branches with spacing that fits them, run with `L` or from the **Layout** menu. See [Organize a graph](https://docs.nodespell.com/graph/organize/). ### Minimap The minimap is a small overview of the whole graph that shows nodes, groups and your current view, opened from **View Settings** in the Graph panel. See [Graph Editor](https://docs.nodespell.com/graph/graph-editor/). ### Bypass Bypass is a node state that skips the node when the graph runs, toggled with `B` on the selected node. See [Graph Editor](https://docs.nodespell.com/graph/graph-editor/). ### Workflow A Workflow is a reusable piece of graph, from two nodes to hundreds, saved with **Save As Workflow** and placed into any Project from the Library. See [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/). ### Node preset A node preset is a named set of parameter values for one node type, such as **Warm Contrast** for Color Correct, that you can create as a new node or apply to selected nodes. See [Snippets and presets](https://docs.nodespell.com/graph/snippets-and-presets/). ## Running and results ### Run A run executes the selected nodes and anything they need upstream, following a plan that decides which nodes generate, reuse or update. See [Run nodes](https://docs.nodespell.com/graph/run-nodes/). ### Run Selected Tasks Run Selected Tasks is the command that runs the selected nodes in Graph, with the shortcut `Cmd/Ctrl+Shift+R`. See [Run nodes](https://docs.nodespell.com/graph/run-nodes/). ### Run plan A run plan is Nodespell's preview of what a run will do to each node, generate, reuse or update, shown when you hover the run button or choose **Plan this run…**. See [Run nodes](https://docs.nodespell.com/graph/run-nodes/). ### Take A take is one saved result of a node's run; every run adds a take and earlier takes are kept, not overwritten. See [Takes](https://docs.nodespell.com/graph/takes/). ### Follow latest Follow latest is the take setting where a node's output always uses its newest successful take. See [Takes](https://docs.nodespell.com/graph/takes/). ### Keep this take Keep this take is the take action that holds a node on one chosen take, so new takes are added to the history without replacing it. See [Takes](https://docs.nodespell.com/graph/takes/). ### Generations Generations is the Create control that sets how many runs Nodespell submits each time you select the action button, from 1 to 4. See [Make your first image](https://docs.nodespell.com/start/first-image/). ### Task A task is one unit of AI work submitted by a run, listed in the Tasks panel with its status, timing and credit change. See [Tasks](https://docs.nodespell.com/graph/tasks/). ### Fullscreen viewer The fullscreen viewer shows a result at full size; open it with `Space` on a selected node or **View fullscreen** in Create, and close it with `Esc`. See [Viewers](https://docs.nodespell.com/edit/viewers/). ## AI models and tools ### Creative Tool A Creative Tool is a Nodespell-authored guided presentation of AI model nodes for a familiar task, such as **Image Generator** or **Image to Video**; it changes the controls you see, not the model, price, results or connections. See [Creative Tools](https://docs.nodespell.com/creative-tools/). ### Model node A model node is a specific AI model shown with its own name and its full set of parameters, rather than through a Creative Tool. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). ### Model picker The model picker is the **Choose model** dialog that lists models for a Creative Tool, with collections such as **Featured**, **Recent**, **Favourites** and **All models**, search and filters. See [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/). ### Branch A branch is a new node Nodespell creates when you change the model of a node that already has takes, so the original node, its takes and its downstream graph stay unchanged. See [Branch and switch presentation](https://docs.nodespell.com/graph/branch-and-switch-presentation/). ### Continue with this result Continue with this result is the Create action that starts **Image Upscaler** or **Image to Video** from an exact chosen image, preserved as a visible Asset node. See [Make your first image](https://docs.nodespell.com/start/first-image/). ### Agent The Agent is Nodespell's in-product assistant, shown in the **Assistant** panel, that can answer questions about the graph and propose or make visible changes to it. See [Agent](https://docs.nodespell.com/agent/). ## Editing and finishing ### Compositor A compositor is a node context for layering and finishing images or video with nodes such as Merge, color correction, masks and transforms, shown in the Design panel. See [Compositing](https://docs.nodespell.com/edit/compositing/). ### Merge Merge is the compositor node that combines two images with a blend mode, or with per-channel expressions in **Custom** mode. See [Compositing](https://docs.nodespell.com/edit/compositing/). ### Keyframe A keyframe is a saved parameter value at a point in time, so the value animates between keyframes. See [Keyframes](https://docs.nodespell.com/edit/keyframes/). ### Timeline The Timeline is the multi-track editor in the Design panel for arranging video and audio clips over time. See [Timeline](https://docs.nodespell.com/edit/timeline/). ## Assets, accounts and teams ### Asset An Asset is a media file stored in Nodespell, uploaded by you or saved from a result, and managed in the Asset Manager. See [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ### Asset Manager The Asset Manager is the **Assets** panel for browsing, searching, tagging, previewing, organizing and reusing Assets. See [Asset Manager](https://docs.nodespell.com/assets/asset-manager/). ### Personal Assets and Team Assets Personal Assets stay in your own library, while Team Assets are shared with authorized members of a Team; moving folders does not change which one an Asset belongs to. See [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/). ### Credits Credits are the Nodespell usage balance that AI model runs draw from; different models and settings use different amounts. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ### Credit estimate A credit estimate is the expected credit cost of a run for the selected model and settings, shown before you run on paid plans. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ### Team A Team is a shared Nodespell workspace with its own Projects, Assets and members; on the Studio plan it also has a shared credit pool. See [Teams](https://docs.nodespell.com/teams/). ## Related - [Core concepts](https://docs.nodespell.com/start/core-concepts/) — the main ideas in one page. - [Panels reference](https://docs.nodespell.com/reference/panels/) — every panel and its purpose. - [Shortcuts](https://docs.nodespell.com/reference/shortcuts/) — keyboard shortcuts. - [FAQ](https://docs.nodespell.com/reference/faq/) — common questions about Nodespell. --- # Troubleshooting > Fix common Nodespell problems by symptom: connections, runs and credits, takes, saving and recovery, uploads, browser loading and the Agent. Source: https://docs.nodespell.com/reference/troubleshooting/ · Last updated: 2026-11-10 This Nodespell troubleshooting guide lists common symptoms, their likely cause and the fix, grouped by area. Messages in bold are quoted exactly as Nodespell shows them, so you can search this page for the text you see. Each fix links to the guide that explains the feature in full. ## Connections | Symptom | Likely cause | Fix | | ---------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------- | | **Check input connections** appears when you run | An input receives a different media type than it expects, for example an image connected to a video input. | Reconnect the asset to a matching input, or choose **Run anyway** if it is intentional. See [Connections](https://docs.nodespell.com/graph/connections/). | | A socket tooltip says **Expects** one type, **got** another | The connected output's content type does not match the input. | Connect an output of the expected type. Hover a socket to see its **Type** and **Content**. | | **Some connections on** a node **could not be matched to its current inputs or outputs** | The node's inputs changed, for example after a model update. | The original connections are preserved; reconnect the affected ones. | | **Not currently driving. A connection is driving this parameter.** | A connection overrides the parameter's expression or value. | Disconnect the input if you want the typed value or [expression](https://docs.nodespell.com/graph/expressions/) to apply. | ## Running and credits | Symptom | Likely cause | Fix | | ---------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------- | | The run control says **Up to date** | Everything selected already has current results. | To make a new result, open **Run options** › **Plan this run…** and set the node to **Generate**. See [Run nodes](https://docs.nodespell.com/graph/run-nodes/). | | A downstream run did not regenerate an upstream image | Runs reuse existing AI takes until you explicitly generate. | Select the upstream node and choose **Generate**. | | **N nodes need a take first** | Upstream AI nodes have no result yet. | Use **Generate** with **& run**, or run those nodes first. | | **Not enough credits for this run.** | Your balance cannot cover the estimate. | Lower the number of takes, or add credits. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). | | **This team's shared credits are empty** | The Team credit pool is used up; personal credits are not used in Team projects. | Ask the Team owner to top up, or wait for the next refresh. See [Team credits](https://docs.nodespell.com/teams/credits-and-usage/). | | No credit estimate next to **Generate** | Live estimates are a paid-plan feature. | You can still run. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). | | The run control shows **Retry loading plan** | The run plan could not be prepared. The tooltip names the reason, such as **A media reference in a prompt is missing.** or **An input is not ready.** | Fix the named cause, then click **Retry loading plan**. | | **upstream asset upload(s) are missing or failed** | An input file did not finish uploading. | Open **Failed uploads** in the top bar, retry, then run again. | | **Run Map body node directly?** | You ran a node inside a Map Container. | Run the Map Container instead. See [Data and logic](https://docs.nodespell.com/graph/data-and-logic/). | | **Folders cannot be run directly.** | A Folder is organization only. | Go inside and run the nodes there. See [Levels and Folders](https://docs.nodespell.com/graph/levels-and-folders/). | | A run in **Task queue** says **Waiting for browser** or **Resume browser session** | Part of the run executes in your browser and the tab was closed or reloaded. | Click **Resume** in **Task queue**. See [Tasks panel](https://docs.nodespell.com/graph/tasks/). | | A task failed | The model rejected the input, or an input was missing. | Read the error in the [Tasks panel](https://docs.nodespell.com/graph/tasks/) or use **Tasks** › **Copy error** on the node, fix the cause, and run again. | ## Results and takes | Symptom | Likely cause | Fix | | ----------------------------------------------------------------------- | -------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------- | | **This take could not be used. The node still uses its previous take.** | The change could not be confirmed, for example during a reconnect. | Try **Use** again once the project is connected. See [Takes](https://docs.nodespell.com/graph/takes/). | | **Use** is unavailable and says **Only successful takes can be used.** | The take failed or was cancelled. | Choose a successful take, or run again. | | **Wait until the take finishes.** | The take is still generating. | Wait for it to complete. | | **You're viewing this project, so takes can't be changed.** | You are viewing, not editing. | Choose **Edit here** or **Request editing**; see [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/). | | A take **is protected** and cannot be deleted | A pinned branch, a saved snapshot, a copy or a running generation uses it. | Remove that reference first, or keep the take. | | A new run replaced the result you liked | The node follows the latest take. | Open **Takes…** and click **Use** on the take you want. The node then keeps it. | | **Some takes couldn't load. The node keeps its current output.** | A connection problem while loading history. | Click **Retry loading**. | ## Saving and recovery | Symptom | Likely cause | Fix | | --------------------------------------- | ------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------- | | The save status shows **Saved locally** | Changes are kept on this device and waiting for the server. | Keep the tab open until it shows saved. | | **Editing paused** | Too many changes are waiting to save (editing pauses at 99). | Wait for saves to finish; check your connection. | | **Needs attention** | Some edits could not be saved. | Open the status to review **Unsaved changes**: use yours, discard them, or **Export for support**. See [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/). | | **Uploads pending** | Media is still uploading and is not in the local backup. | Keep the tab open until uploads finish. | | **Your other tab isn't responding** | Another tab or device holds editing for this project. | Choose **Keep waiting**, or **Take over now**. | ## Assets and uploads | Symptom | Likely cause | Fix | | --------------------------------------- | ---------------------------------------------------------------- | --------------------------------------------------------------------------------------- | | **Unsupported file types** | The file format is not accepted. | Convert it to a supported format. See [Supported media](https://docs.nodespell.com/reference/supported-media/). | | **Files too big** or **Too many files** | The upload is over the size or count limit shown in the message. | Upload fewer or smaller files. See [Upload and organize](https://docs.nodespell.com/assets/upload-and-organize/). | | **Assets could not be loaded.** | A connection problem. | Check your connection and reopen the Asset Manager. | ## Browser and loading | Symptom | Likely cause | Fix | | ------------------------------------------------------------------ | ----------------------------------------------------------------------- | ------------------------------------------------------------------------ | | **Nodespell could not load. Check your connection and try again.** | The app could not load. | Click **Reload**. | | **Nodespell has been updated** | A newer version is available. | Click **Reload Nodespell** to continue where you left off. | | **This panel ran into a problem** | One panel failed; the rest of the workspace is unaffected. | Click **Try again**, or **Report this problem** with the copied details. | | **Preview not supported in this browser** | The browser cannot show that preview. | Try another browser. | | Keyboard shortcuts do nothing | The graph panel does not have focus, or a text field or menu is active. | Click the graph, then try again. See [Shortcuts](https://docs.nodespell.com/reference/shortcuts/). | ## Agent | Symptom | Likely cause | Fix | | ----------------------------------------------------------------- | --------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------- | | **Open a graph panel first.** | The Agent needs a graph to work on. | Open a Graph panel, then send the request again. See [Agent](https://docs.nodespell.com/agent/). | | **Another assistant request is already running in this project.** | One Agent request runs per project at a time. | Stop it or wait for it to finish. | | **Turn budget reached** | The turn hit its spending limit. | Check **Spending limit per turn** in the Agent composer, then continue. See [Sessions and models](https://docs.nodespell.com/agent/sessions-and-models/). | | **The assistant paused after repeated failures.** | The same step failed several times. | Edit the prompt before trying again. | | **This proposal is out of date.** | The graph changed after the Agent proposed a change. | Ask the Agent to review the current graph. | | An attachment is rejected | Over 12 files at once, over 200MB each, or an unsupported type. | Attach fewer or smaller files. See [Context and references](https://docs.nodespell.com/agent/context-and-references/). | ## Frequently asked questions ### What should I include when I contact support? Include the task ID (**Copy Task ID** in the [Tasks panel](https://docs.nodespell.com/graph/tasks/)), the exact error message, and any support code shown. See [Support](https://docs.nodespell.com/account/support/). ### Why did my run not spend credits? Nodespell reused existing takes. Only nodes marked **Generate** in the run plan spend credits. ### Will I lose work if saving fails? Unsaved edits are kept on your device and listed for review, and you can export a copy. Keep the tab open until the status shows saved. ## Related - [Run nodes](https://docs.nodespell.com/graph/run-nodes/): run plans, costs and stopping runs - [Tasks panel](https://docs.nodespell.com/graph/tasks/): failure details and usage - [Takes](https://docs.nodespell.com/graph/takes/): choosing and deleting results - [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/): unsaved edits in depth - [Support](https://docs.nodespell.com/account/support/): getting help --- # Nodespell frequently asked questions > Answers to common questions about Nodespell, including what it is, browser needs, where work runs and is saved, credits, plans, teams and getting help. Source: https://docs.nodespell.com/reference/faq/ · Last updated: 2026-11-10 The Nodespell FAQ answers general questions about Nodespell: what it is, what you need to use it, where your work runs and is saved, how credits and plans work, and where to get help. Each answer stands on its own and links to the guide with the details. | Topic | Questions | | ----------------- | -------------------------------------------------------- | | The product | What Nodespell is, who it is for, which models it offers | | Requirements | Browser, installation, window size, age, languages | | Your work | Where it runs, where it is saved, what happens offline | | Credits and plans | What credits are, the free plan, video, running out | | Teams and help | Working with others, the Agent, getting support | ## Frequently asked questions ### What is Nodespell? Nodespell is a browser-based creative workspace where you generate and finish images, video, audio and text by connecting nodes in a graph. AI models, edits and effects are nodes, and each result is kept as a take on the node that made it. See [What is Nodespell?](https://docs.nodespell.com/start/what-is-nodespell/). ### Who is Nodespell for? Nodespell is for artists, designers, motion and VFX artists and studios who want control over each step of AI-assisted work. Beginners can start in the guided **Create** workspace, and people used to node editors can work directly in **Graph**. See [Choose a workspace mode](https://docs.nodespell.com/start/workspace-modes/). ### Which AI models does Nodespell offer? Nodespell offers image, video, audio and text models from several model makers, presented as model nodes and through Creative Tools. The list changes as models are added and retired; the model picker in the app shows what is available now, and [What's new](https://docs.nodespell.com/whats-new/) records each addition. ### Do I need to install anything? No. Nodespell runs in a web browser at app.nodespell.com. AI model nodes run on Nodespell's servers, so you do not need a local GPU or model downloads. ### Which browser should I use? Use an up-to-date desktop browser. Some features check that your browser has the capabilities they need: the Timeline editor asks for an up-to-date Chromium- or Firefox-based desktop browser, and exporting compositor image sequences asks for desktop Chrome or Edge. ### What screen size does Nodespell need? The **Create** workspace needs a window at least 1080 pixels wide; in a narrower window it offers **Open Graph** instead. **Graph** and **Immersive** have no stated minimum width. ### How old do I have to be to use Nodespell? When you sign up, you confirm that you are at least 18, agree to the Terms and Conditions and acknowledge the Privacy Notice. Both documents are linked from the sign-up page and from **Help** > **About**. ### Which languages is the Nodespell interface available in? The interface is available in English, UK English and Simplified Chinese. Change it in Preferences or with the command palette. See [Account settings](https://docs.nodespell.com/account/settings/). ### Where does my work run? AI model nodes run on Nodespell's servers and use credits. Compositor and image-processing nodes render in your browser using your computer's graphics hardware. Either way, results are saved to your Project. ### Where is my work saved? Nodespell saves your Projects, takes and uploaded Assets to your account automatically; the top bar shows **Saved** when your latest changes are stored. If saving is interrupted, Nodespell keeps a backup in the browser on your device and retries. See [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/). ### What happens if I lose my internet connection? Nodespell shows **You're offline** and some features stop working until you reconnect. Unsaved changes are kept on your device and saved when the connection returns; keep the tab open until the top bar shows **Saved**. AI model runs need a connection. ### What are credits? Credits are the usage balance for running AI models in Nodespell. Different models use different amounts depending on the model, its settings and the output length. Opening Projects, editing and running with existing takes does not use credits. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ### Is there a free plan? Yes. Nodespell Free lets you explore Nodespell, open Projects and work with your own media. AI generation uses credits, credits on a free account do not renew, and video generation is part of paid plans. See [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/). ### Can I see what a run will cost before I start it? On paid plans, Nodespell shows a credit estimate for the selected model and settings before you run, and updates it as you change settings. Free accounts see a locked **Cost estimate** slot instead. ### What happens when I run out of credits? Your Projects, Assets and workflows stay intact, and you can still review and edit your work. New AI model runs pause until more credits are available. Paid plans include monthly credits, and paid users can buy more. ### Can I use my own images, video and audio? Yes. Upload media to the Asset Manager or drag files onto the graph, then use them as inputs to any compatible node. Formats and sizes are listed in [Supported media](https://docs.nodespell.com/reference/supported-media/). ### Can I work with other people? Yes. A Team is a shared workspace with its own Projects, Assets and members. On the Studio plan, a Team also has a shared credit pool managed by the team owner. See [Teams](https://docs.nodespell.com/teams/). ### Does the Agent use credits? The Agent's usage can be charged in credits, and the **Assistant** panel shows the credits used by the last turn and the session. Any generation the Agent starts is a normal run with its own cost. See [Agent](https://docs.nodespell.com/agent/). ### How do I get help or report a problem? Open the **Help** menu. **Help** opens this documentation, **Report a problem…** prepares a support message with technical details, **Email** copies the support address, and **Discord** opens the community. See [Support](https://docs.nodespell.com/account/support/). ### How do I find out what changed in Nodespell? Choose **Help** > **What's new** in the app, or **More** > **What's new** on the dashboard. The same release notes are on the [What's new](https://docs.nodespell.com/whats-new/) page. ## Related - [What is Nodespell?](https://docs.nodespell.com/start/what-is-nodespell/) — a full overview. - [Core concepts](https://docs.nodespell.com/start/core-concepts/) — Projects, nodes, takes and credits. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/) — plans, balances and estimates. - [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/) — symptoms, causes and fixes. - [Glossary](https://docs.nodespell.com/reference/glossary/) — every Nodespell term. --- # Nodespell plans, credits and cost estimates > Compare Nodespell plans and their limits, learn what credits are and what uses them, and read the credit estimates shown before you run a model. Source: https://docs.nodespell.com/account/plans-and-credits/ · Last updated: 2026-11-10 Credits are the usage balance Nodespell uses for running AI models, including model nodes, Creative Tools and the Agent. Your plan sets how many credits you receive each month and your limits for projects, storage, team members and retention. Different models use different amounts of credits depending on the model, its settings, inputs and output length, and paid plans show an estimate before you run. ## Which plans are available? Nodespell has four plan levels. Starter, Pro and Pro Plus are personal plans. Studio is a Team plan billed to a Team owner. | | Free | Starter | Pro | Pro Plus | Studio | | ----------------------- | ------ | -------- | --------- | --------- | ------------------------------------------ | | Monthly credits | None | 13,000 | 30,000 | 60,000 | 32,000 per team member, in one shared pool | | Active projects | 3 | 120 | Unlimited | Unlimited | Unlimited | | Storage | 250 MB | 10 GB | 25 GB | 25 GB | 50 GB | | CDN bandwidth per month | 250 MB | 10 GB | 25 GB | 25 GB | 50 GB | | Team members | None | None | Up to 3 | Up to 3 | 2 to 10 | | Snapshot retention | — | 30 days | 1 year | 1 year | 2 years | | Asset retention | — | 3 months | 1 year | 1 year | 2 years | | Credit top-ups | No | Yes | Yes | Yes | Yes, to the Team pool | | Video generation | No | Yes | Yes | Yes | Yes | | Commercial use | No | Yes | Yes | Yes | Yes | Monthly plan credits refresh at the start of each billing cycle and do not roll over. Plans can be billed monthly or yearly. Prices are shown in the plan picker and at checkout, where taxes are calculated. Studio checkout covers up to 4 team members; larger Studio Teams are set up with Nodespell. To compare plans in the app, open **Subscription** in the dashboard sidebar, or **Account Settings** > **Subscription**, and open the plan picker. **Compare all features** shows the full comparison, and **See what your credits can make** shows sample credit use for popular models. ## What uses credits? | Uses credits | Does not use credits | | ---------------------------------------------------------- | ----------------------------------------------------- | | Running AI model nodes in Graph | Building and editing graphs | | Generating with Creative Tools in Create | Uploading and organizing Assets | | Agent replies, charged per turn | Non-AI nodes such as compositing, trimming and export | | Generations the Agent runs for you, after you approve them | Browsing and viewing results | The Agent's own replies use a small amount of credits for each turn, and the Agent asks before it spends credits on a generation. See [Sessions and models](https://docs.nodespell.com/agent/sessions-and-models/). When a run starts, its credits are held from your available balance. You are charged only for completed work, and unused held credits are released. ## Read the credit estimate before you run On paid plans, run controls and menus show a credit estimate that updates as you change settings. | Estimate | Meaning | | ----------- | -------------------------------------------------------------- | | **~120** | Roughly that many credits. | | **120+** | At least that many credits; some parts could not be estimated. | | **Varies** | The cost depends on input size. | | **Unknown** | No estimate is available. | The estimate's details say how confident it is (**High-confidence estimate**, **Estimated**, **Typical**, **Partial estimate**) and why, for example **Input-size dependent** or **AI cache may reduce actual spend**. If a run may cost more than you have, the details say **This run may cost more than your balance.** You can hide estimates with **Show Execution Credit Estimates** in [Preferences](https://docs.nodespell.com/account/settings/). Estimates are not prices. The final charge depends on the work actually completed. ## Why is the cost estimate locked? On the Free plan, the estimate slot shows a locked **Cost estimate** chip. Selecting it explains that paid plans show a live credit estimate there. You can still run models your plan allows. In a Team workspace, estimates follow the Team's plan, and the chip suggests asking the Team owner to upgrade. ## Where to see your credits The credit display in the top bar shows your remaining credits for the current workspace, how much of your allowance is used, and when credits renew or expire. For full details open **Account Settings** > **Subscription**. | Balance | What it holds | | ------------------------ | -------------------------------------------------------------------------- | | **Subscription credits** | Your plan's monthly credits. They refresh each cycle and do not roll over. | | **Top-up credits** | Credits you bought. They expire one year after purchase. | | **Other credits** | Trial, referral and other one-time grants. | **Credit details** lists each **Active credit grant** with its source and how much is left. **Refresh credits** reloads the balance. ## Free accounts and trial credits A Free account can explore Nodespell, open projects and work with your own media. AI generation uses credits, and video generation is part of paid plans. New accounts may receive one-time free credits after email verification; they do not renew. Referral rewards from **Refer and Earn** are also added as credits. ## What happens when you reach a limit? Your plan usage is shown under **Plan usage** in **Account Settings** > **Subscription**, with **Active projects**, **Storage**, **CDN bandwidth**, **Team members**, **Snapshot retention** and **Asset retention**. Each shows **Healthy**, **Approaching limit** or **Limit reached**. When an action goes over a limit, Nodespell shows **Plan limit reached** with **View plans**. Retention cleanup uses creation time and keeps Assets referenced by active projects protected. ## Frequently asked questions ### What happens if I run out of credits? Your projects, Assets and Workflows stay intact, and you can still review and edit your work. New AI runs pause until more credits are available from a top-up or your next monthly refresh. ### Do unused monthly credits roll over? No. Monthly plan credits refresh at the start of each billing cycle and do not roll over. Purchased top-up credits last one year from purchase. ### Does the Agent use credits? Yes. Each Agent reply uses credits, and any generation the Agent runs uses credits like a normal run. The Agent asks before it spends credits on generation. ### Why can I not generate video on the Free plan? Video generation is part of paid plans. Upgrade to Starter or higher to run video models. ### Where can I see what each model costs? Paid plans show a live estimate in the run controls before each run. The plan comparison also shows sample credit use for popular models at fixed settings. ## Related - [Billing](https://docs.nodespell.com/account/billing/): subscribe, change plans and buy credits. - [Team credits and usage](https://docs.nodespell.com/teams/credits-and-usage/): shared Team pools. - [Run nodes](https://docs.nodespell.com/graph/run-nodes/): estimates in Graph. - [Choose a model](https://docs.nodespell.com/creative-tools/choose-a-model/): estimates in Create. - [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/): storage limits. --- # Manage your subscription, invoices and credit top-ups > Subscribe to a Nodespell plan, change or stop it, open invoices, buy credit top-ups, and learn what stays available in the 14-day grace period. Source: https://docs.nodespell.com/account/billing/ · Last updated: 2026-11-10 Nodespell billing covers your subscription plan, its invoices and payment method, and any extra credits you buy as top-ups. You manage all of it from **Account Settings** > **Subscription**. Payments go through a secure Stripe checkout. When a paid plan ends, its limits stay for 14 days, and credits you bought stay usable until they expire. ## Open your billing settings Open **Subscription** in the dashboard sidebar, or choose **My account** from the account menu and open the **Subscription** tab. In a Team workspace the sidebar item is **Team billing**, and the tab shows **Personal workspace** and **Team billing** views. The tab shows your current plan, when it started and **Renews**, your **Available Credits**, and buttons such as **Change plan**, **Billing & invoices** and **Buy credits**. ## Subscribe or upgrade 1. In **Subscription**, open the plan picker (**Choose your plan**). 2. Choose **Monthly** or **Yearly** billing. 3. Select the button on the plan you want, such as **Get Started**. Plans that need approval show **Request access** or **Contact us** instead. 4. Complete the Stripe checkout. Taxes are calculated at checkout. If you already have a paid plan, choosing a higher plan opens **Review subscription change**. It shows **Current billing**, **New billing**, **Due today** and **Next renewal**. Select **Confirm change** to upgrade now; Stripe charges the saved payment method, and any prorated amount and tax are calculated when you confirm. ## Downgrade or switch billing cycle Downgrades, and switches from yearly to monthly billing, take effect at your next renewal so your current access and credits stay the same until then. The review dialog says **Downgrade at renewal** or **Switch at renewal**, and **Schedule change** confirms it. If you already have a scheduled change, a new one replaces it after you confirm **Replace scheduled change**. ## Stop your paid plan at renewal 1. Open the plan picker and scroll to the bottom. 2. Under **Manage subscription**, select **Stop paid plan at renewal**. 3. In **Review subscription end**, select **Stop at renewal**. Your paid access and credits stay active until the end of the current billing cycle, and no further recurring charge is created. The plan then shows **Scheduled cancellation** with the end date. To keep a paid plan, choose another paid plan before renewal. ## View invoices and update your payment method Select **Billing & invoices** to open the billing portal, where you can update your payment method and billing information and see invoices and receipts. If you have not bought anything yet, Nodespell shows **No billing history yet**. If a paid subscription you bought is not showing, the Free plan card can offer **Restore Subscriptions** to look it up again. ## Buy extra credits Top-ups add prepaid credits on top of your plan. They are available on Starter, Pro and Studio, and the Free plan must upgrade first. 1. Select **Buy credits** in **Subscription**. 2. Choose a credit pack and a **Quantity**. Nodespell calculates your credit quote before checkout. 3. Select **Continue to checkout** and complete payment. | Top-up type | Where the credits go | | --------------------- | ------------------------------------------------------------ | | **Personal top-ups** | Your personal workspace only. Team projects do not use them. | | **Team pool top-ups** | The shared pool of a Studio Team you own. | Top-up spend has a **Monthly top-up allowance** based on your paid subscription history. For more than the allowance or the available packs, use **Contact us**. Purchased credits expire one year after purchase. ## What happens when your paid plan ends? Your previous plan's limits stay for **14 days** after your paid service period ends. During this grace period you can keep using credits that are still valid, but no new monthly credits are added and credit expiry dates do not change. After the grace period, valid purchased credits keep basic prepaid access until they are used or expire: | Credits purchased | Storage | Active projects | | -------------------------------------- | ------- | --------------- | | Before 3 November 2026, 00:00 UTC | 10 GiB | 120 | | On or after 3 November 2026, 00:00 UTC | 5 GiB | 10 | A new purchase does not extend an older allowance, and each purchase keeps its one-year expiry. Prepaid access includes credit-funded generation and commercial use of those outputs, subject to the applicable model terms. It does not include new monthly credits, new collaborator invitations, premium support or concurrency, or extra CDN allowance. Existing projects stay editable even above the project limit, but you must archive projects before creating or reactivating more, and free storage before adding files. Nodespell does not archive or delete your work because your plan ends; normal retention rules still apply. Existing Studio members keep using the same Team pool, and personal and Team credits are not combined. ## If a renewal payment fails The **Subscription** tab shows that your renewal payment was not completed and that your previous plan access continues until a date. Review your billing and the billing emails you received. In some cases the app asks for a billing review before another payment, to avoid charging twice; contact info@nodespell.com when it does. ## Studio billing for Teams Studio is billed to the owner of a Team. In **Team billing**, the **Studio team plan** card lets you choose the number of team members, starting at 2. Nodespell creates a billing Team if you do not own one yet. Your personal plan and credits stay active and are billed separately. To change the number of paid team members later, contact support. ## Frequently asked questions ### Do I keep my credits if I cancel? Your paid access and credits stay until the end of the billing cycle. After that, plan limits continue for 14 days, and valid purchased top-up credits stay usable until they expire. ### Do you offer refunds? The app states that all purchases are final and refunds are not offered at this time. Contact support if you have a problem with your subscription. ### When does a downgrade take effect? At your next renewal date. Your current access and credits stay the same until then. ### Can I buy credits on the Free plan? No. Top-ups are available after upgrading to Starter, Pro or Studio. ### Where do I get a copy of an invoice? Select **Billing & invoices** in **Subscription**. Contact support if you need a copy that is not there. ## Related - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): compare plans and limits. - [Team credits and usage](https://docs.nodespell.com/teams/credits-and-usage/): Studio pools. - [Account settings](https://docs.nodespell.com/account/settings/): other account tabs. - [Support](https://docs.nodespell.com/account/support/): contact Nodespell. - [Workspaces and storage](https://docs.nodespell.com/assets/workspaces-and-storage/): storage limit messages. --- # Account settings and Preferences > Change your Nodespell profile, creator profile and notifications in Account Settings, and set theme, accent, language and saved text in Preferences. Source: https://docs.nodespell.com/account/settings/ · Last updated: 2026-11-10 Nodespell has two places for settings. **Account Settings** holds your profile, public creator profile, subscription, referrals, notifications, Teams and security. **Preferences** controls how the app looks and behaves, including theme, accent colour, language, saved text and editor defaults. Preferences are saved to your account, so they follow you when you sign in on another device. ## Open Account Settings Choose **My account** from the account menu, at the bottom of the dashboard sidebar or in the editor's top bar. Account Settings opens as a window with these tabs: | Tab | What you can change | | ------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------- | | **Profile** | **First Name**, **Last Name** and **Avatar** (JPG, PNG or WebP, up to 5 MB). Your **Email** is shown but cannot be changed after registration. | | **Creator Profile** | Your public display name, creator URL, bio, location, links and whether your email is public. | | **Subscription** | Your plan, credits, invoices and top-ups. See [Billing](https://docs.nodespell.com/account/billing/). | | **Referrals** | Your referral code and link, and referral activity. | | **Notification** | Whether the **Continuous Improvement Survey** appears in the app. | | **Teams** | Your Teams, invitations, members and Team credit usage. See [Teams](https://docs.nodespell.com/teams/). | | **Security** | Multi-factor authentication, password, legal documents and account deletion. See [Sign-in and security](https://docs.nodespell.com/account/sign-in-and-security/). | After changing fields on **Profile**, **Creator Profile** or **Notification**, select **Save Changes**. ## Set up your public creator profile Your creator profile appears on Spellworks, Workflows and Projects you publish. | Field | Details | | -------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Public display name** | The name shown on your profile and published work. | | **Creator URL** | Your public profile address. You can change it once every 30 days, and old URLs redirect to the current one. Use lowercase letters, numbers and hyphens. | | **Creator bio** | Up to 500 characters. | | **Location** | Where you or your studio are based. | | **Creator social links** | Up to 8 links. Paste a URL and Nodespell detects the type; **Website** accepts any domain. | | **Show email in public profile** | Off by default. Turn it on to show your email on your public profile. | ## Open Preferences Choose **Edit** > **Preferences** in the editor, or press `Cmd+,` on macOS or `Ctrl+,` on Windows. Use **Search all settings...** to find a setting by name, description or option. **Reset to Defaults** restores every preference after you confirm **Reset preferences**, and each setting has its own **Reset to default**. Preferences are grouped into three sections: | Group | Categories | | ---------------- | ------------------------------------------------------------------- | | **App** | **General**, **User Interface**, **Hotkeys**, **Saved text** | | **Workspace** | **Workspace Modes**, **Node Graph Editor**, **Panels**, **Viewers** | | **Data & tools** | **My node defaults**, **Advanced** | ## Change the theme and accent colour Open **Preferences** > **User Interface**. | Setting | Options | | -------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------- | | **Theme** | **Dark** (default), **Light**, or **System** to follow your operating system. | | **Interface Style** | **Classic** (default), a softer look with rounded panels, or **Machinery**, an angular look with its own typeface. | | **UI Text Size** | **Smallest**, **Smaller**, **Regular** (default), **Larger**, **Largest**. Scales text without changing layout or graph zoom. | | **Accent Palette** | **Nodespell** (default), 17 named colours from **Red** to **Rose**, or **Custom** with a **Custom Accent Color**. | | **Dark Mode Background** / **Light Mode Background** | Brightness of the interface background, 0 to 100. | | **Background Tint Color** / **Background Tint Strength** | A subtle colour tint for neutral surfaces. Strength 0 means no tint. | | **Graph Background** | **Auto** keeps the graph canvas slightly darker than the interface; **Custom** uses the **Graph Background Level** you set. | | **Show Credit Balance In Top Bar** | Shows or hides the credit display. | | **Show Status Bar Hotkey Help** | Shows or hides shortcut hints in the status bar. | ## Change the language Open **Preferences** > **General** and choose a **Language**: **English (US)**, **English (UK)** or **Chinese Simplified**. You can also change it from **More** > **Language** in the dashboard sidebar. **General** also has **Input Device Type**: **Auto**, **Mouse** or **Touchpad**, which tunes canvas navigation. ## Choose which workspace opens In **Preferences** > **Workspace Modes**, **Default Workspace** sets whether Projects open in **Create**, **Graph** or **Immersive**. You can still switch from the workspace header. See [Workspace modes](https://docs.nodespell.com/start/workspace-modes/). ## Manage saved text Saved text is your personal library of reusable prompts and text. Open **Preferences** > **Saved text** to add, edit, search, **Pin to top** or delete entries. Sort by **Recently edited**, **Name** or **Recently used on this device**. 1. Select **New**, add the **Text** and an optional **Name**, then select **Save**. 2. In any text field, type `/` or right-click and choose **Insert saved text**. 3. To save text you have selected in a field, right-click and choose **Save selection as text…**. Saved text can be up to 20,000 characters, with names up to 200 characters. It syncs across your devices. When you are offline, changes are kept on the device and saved when you reconnect. If the same entry changes on two devices, Nodespell asks which version to keep. See [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/). ## Other Preferences - **Hotkeys** lists every keyboard shortcut with its panel and function. Search by name or key. See [Shortcuts](https://docs.nodespell.com/reference/shortcuts/). - **Node Graph Editor**, **Panels** and **Viewers** set editor behaviour, such as grid, minimap, connection style and fullscreen playback. - **My node defaults** holds the settings used when you add a node of each type. - **Advanced** holds maintenance actions, such as restoring the Create draft warning. ## Cookie settings Choose **Cookie settings** from the account menu to change which optional cookies Nodespell may use in this browser: **Necessary** are always on; **Analytics** and **Marketing** are optional. ## Frequently asked questions ### Can I change my email address? No. The **Profile** tab notes that your email cannot be changed after registration. ### Do my Preferences sync between devices? Yes. Preferences are saved to your account. Saved text also syncs across devices. ### Can I change keyboard shortcuts? The **Hotkeys** category lists and searches shortcuts. It does not offer a way to reassign them. ### How do I make my email public? Open **Account Settings** > **Creator Profile**, turn on **Show email in public profile**, then select **Save Changes**. ## Related - [Sign-in and security](https://docs.nodespell.com/account/sign-in-and-security/): password, MFA and account deletion. - [Billing](https://docs.nodespell.com/account/billing/): the Subscription tab. - [Prompts and saved text](https://docs.nodespell.com/graph/prompts-and-saved-text/): inserting saved text. - [Interface tour](https://docs.nodespell.com/start/interface/): the workspace layout. - [Shortcuts](https://docs.nodespell.com/reference/shortcuts/): every keyboard shortcut. --- # Sign in, verify your email and secure your account > Sign up and sign in to Nodespell with Google, GitHub, an email link or a password, reset your password, enable MFA, sign out, or delete your account. Source: https://docs.nodespell.com/account/sign-in-and-security/ · Last updated: 2026-11-10 Nodespell accounts sign in with Google, GitHub, a one-time email link, or an email address and password. Each account is tied to one email address, which must be verified before you can use features such as free trial credits and Team invitations. Security settings let you change your password, add an authenticator app for sensitive actions, and delete your account. ## Which sign-in methods are available? | Method | How it works | | -------------------------- | ---------------------------------------------------------------------------------------- | | **Continue with Google** | Signs in with your Google account. | | **Continue with GitHub** | Signs in with your GitHub account. | | **Continue with email** | Emails you a secure sign-in link. No password needed. The link expires after 15 minutes. | | **Use email and password** | Signs in with your email address and a Nodespell password. | The same options create a new account if you do not have one. Signing up confirms that you are at least 18, agree to the Terms and Conditions, and acknowledge the Privacy Notice. ## Create an account with an email and password 1. On the sign-in screen, choose **Use email and password**, then **Sign up**. 2. In **Create your account**, enter **First Name**, **Last Name**, **Email**, **Password** and **Confirm Password**. Add a referral code under **Have a referral code?** if you have one. 3. Select **Create Account**. 4. Open the verification email and select the link. The app shows **Email Verified Successfully**. Passwords must be 16 to 48 characters and include at least one uppercase letter, one lowercase letter, one number and one special character. If the verification email does not arrive, choose **Resend Verification Email**. A link that has expired shows **Verification Link Expired**; request a new one. Accounts created with Google, GitHub or an email link do not need a separate verification step. ## Sign in with an email link 1. Choose **Continue with email**. 2. Enter your **Email** and select **Email me a sign-in link**. 3. Open the email and select the link within 15 minutes. If the link has expired, choose **Request New Magic Link**. **Use a password instead** switches to password sign-in. ## Reset a forgotten password 1. On the password sign-in form, select **Forgot password?**. 2. Enter your **Email** and select **Send Reset Instructions**. 3. Open the email and select the reset link. 4. Enter a **New Password**, confirm it, and select **Reset Password**. A reset link that has already been used shows **Invalid Reset Link**. Request a new one. ## Change your password Open **Account Settings** > **Security** and select the **Change Password** button. A new browser tab opens where you enter your **Current Password**, a **New Password** and **Confirm New Password**, then select **Change Password**. You can also sign in with an email link at any time without a password. ## Protect sensitive actions with multi-factor authentication Multi-factor authentication (MFA) in Nodespell protects sensitive exports and administrative actions with a six-digit code from an authenticator app. It does not add a step to ordinary sign-in. 1. Open **Account Settings** > **Security**, then select **Enable authenticator** under **Multi-factor authentication**. 2. In any TOTP-compatible authenticator app, choose manual entry and enter the **Manual entry key**. 3. Type the **Six-digit code** the app shows and select **Confirm authenticator**. 4. Save the recovery codes Nodespell shows (**Copy codes** or **Download**), then select **I saved them**. Each recovery code works once if your authenticator is unavailable. The codes are not shown again. From the same section you can create **New recovery codes**, **Replace authenticator** or **Disable MFA**; each needs a current code. For your protection, start setup shortly after signing in. ## Sign out Open the account menu and choose **Log out**. ## Delete your account 1. Download anything you need. After deletion you cannot access projects, Assets, API keys or collaboration. 2. If you own a Team with other members, transfer or close it first. 3. Open **Account Settings** > **Security** and select **Delete Account**. 4. Type `DELETE` and select **Confirm**. Nodespell stops account access and future subscription renewal immediately, then aims to delete or anonymise personal data in its active systems within 30 days. Some billing, security, legal and shared-workspace records may be kept as described in the Privacy Notice. > **Caution:** Account deletion cannot be cancelled, and there is no way to restore the account afterward. ## Frequently asked questions ### Can I change the email address on my account? No. The email address cannot be changed after registration. ### How long does a sign-in link last? Fifteen minutes. Request a new link if it expires. ### Does MFA protect my sign-in? No. Nodespell MFA protects sensitive exports and administrative actions. Sign-in uses your chosen method. ### Why can I not delete my account? If you own shared Teams or organisations, Nodespell asks you to transfer or close them before deleting your account. ### Do I need to verify my email? Yes, for email and password accounts. Verification is needed before you can open Team invitations, and any free credits are added after verification. ## Related - [Account settings](https://docs.nodespell.com/account/settings/): profile and Preferences. - [Members and roles](https://docs.nodespell.com/teams/members-and-roles/): accepting Team invitations. - [Billing](https://docs.nodespell.com/account/billing/): what happens to your subscription. - [Plans and credits](https://docs.nodespell.com/account/plans-and-credits/): free trial credits. - [Support](https://docs.nodespell.com/account/support/): contact Nodespell. --- # Get help and contact Nodespell support > Find help in Nodespell from the Help menu and Help and about window, read What's new, join Discord, and contact support with the right details. Source: https://docs.nodespell.com/account/support/ · Last updated: 2026-11-10 Nodespell support is reached from inside the app: the **Help** menu in the editor, the **Help and about** window, and the **More** menu on the projects dashboard. From there you can open these docs, read What's new, report a problem with technical details attached, send feedback, copy the support email address, or join the Nodespell community on Discord and YouTube. ## What is in the Help menu? Open **Help** in the editor's top bar. A dot on the menu means there are updates you have not seen. | Item | What it does | | ----------------------------- | -------------------------------------------------------------- | | **Help** | Opens this documentation site. | | **What's new** | Opens the list of recent updates. | | **Getting started tutorials** | Opens short walkthroughs for your first projects. | | **Show first-run guide** | Shows the first-run guide again, where Create is available. | | **Show suggestions** | Turns in-app suggestions on or off. | | **Email** | Copies the support email address to your clipboard. | | **Report a problem…** | Opens a message to Nodespell with technical details filled in. | | **Discord** | Opens the Nodespell Discord community. | | **YouTube channel** | Opens Nodespell tutorials and demos. | | **Submit feedback** | Opens the feedback form. | | **About** | Opens the **Help and about** window. | On the projects dashboard, the same essentials are under **More** in the sidebar: **What's new**, **Language**, **Copy support email**, **Discord** and **YouTube channel**. ## Use the Help and about window Choose **Help** > **About** to open **Help and about**. It has three sections: - **Get help**: **Documentation**, **Getting started tutorials**, **What's new** (with the latest update date), **Report a problem** and **Send feedback**. - **Community**: Discord, YouTube and the Nodespell website. - **Support details**: your **Build**, **Browser**, **Graphics**, **Language** and **Window** size, with **Copy details**. The footer shows the support email address and links to the Terms and Privacy Notice. ## Read What's new **What's new** lists recent improvements, new features and fixes, newest first, with a marker for updates you have not seen. Open it from **Help** > **What's new**, **More** > **What's new** on the dashboard, or the **Help and about** window. The full history is also at [What's new](https://docs.nodespell.com/whats-new/). ## Report a problem 1. Choose **Help** > **Report a problem…**, or **Report a problem** in **Help and about**. Error screens also offer **Report this problem**. 2. The **Contact Nodespell** dialog opens with the support **Email**, a **Subject**, and a **Message** that already contains technical details. 3. Add what you were doing when the problem happened, above the generated details. 4. Select **Open email app** to send it from your mail app. If nothing opens, use **Copy email** and **Copy request** and paste them into any email. Nothing is sent until you send the email yourself. ## Send feedback or ask a question Choose **Help** > **Submit feedback**. In **Submit Feedback**, pick a **Type** (**Feature Request**, **Bug Report**, **General Feedback** or **Question**), a **Category** where asked (such as **AI Models**, **Project Management** or **Billing & Credits**), and write a **Message** of at least 10 characters. Select **Submit**. ## How to contact support | Channel | Where to find it | | --------------------------------------------------------------------- | ----------------------------------------------------------------------------------------- | | Email: info@nodespell.com | **Help** > **Email**, **More** > **Copy support email**, or the **Help and about** footer | | Discord: [discord.gg/MTv95R93Xx](https://discord.gg/MTv95R93Xx) | **Help** > **Discord** or **More** > **Discord** | | YouTube: [youtube.com/@Nodespell](https://www.youtube.com/@Nodespell) | **Help** > **YouTube channel** or **More** > **YouTube channel** | | Website: [nodespell.com](https://nodespell.com) | **More** > **Visit website** | ## What details should you include? Include enough for Nodespell to find and reproduce the problem: - **Support details** from **Help and about**: select **Copy details** and paste them. They contain your build, browser and system, graphics card, language and window size. - What you were doing, what you expected, and what happened instead. - The Project name, and the node or Creative Tool involved. - The time it happened, and any **Error code** or **Reference** shown in the message. - For edits that could not save, the recovery file from **Export for support** or **Export copy**. See [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/). - For billing questions, the email address on your account. Never send passwords or card numbers. Some error messages, such as a failed version save, include a **Contact us** button that opens the feedback form with the action, problem, reference and time already filled in. You can edit these before you submit. ## Frequently asked questions ### What is the Nodespell support email address? info@nodespell.com. Copy it from **Help** > **Email** or **More** > **Copy support email**. ### Does Report a problem send data automatically? No. It prepares an email with technical details. Nothing is sent until you send it from your email app. ### Where can I ask other Nodespell users for help? Join the Nodespell Discord from **Help** > **Discord** to ask questions and share work. ### Where do I find the latest changes? Open **Help** > **What's new** in the editor, or **More** > **What's new** on the dashboard. ## Related - [Saving and recovery](https://docs.nodespell.com/projects/saving-and-recovery/): exporting unsaved edits for support. - [Troubleshooting](https://docs.nodespell.com/reference/troubleshooting/): fixes for common problems. - [FAQ](https://docs.nodespell.com/reference/faq/): general questions. - [Billing](https://docs.nodespell.com/account/billing/): subscription and invoice help. - [Interface tour](https://docs.nodespell.com/start/interface/): where the Help menu is.