Nodaro DocsCompose Video
Describe the video you want, and an AI model arranges your connected images, clips and audio into a timed, multi-track composition that Render Video renders.
The Compose Video node turns a text prompt and your connected media into a video composition. A language model reads your prompt and the connected images, clips and audio, and arranges them into a multi-track timeline with timing, layers and transitions. The output is a plan, and a Render Video node renders it into the final video.
When to use it
- Assemble generated images and a narration into a slideshow or an explainer video.
- Build a documentary-style piece from several media sources.
- Cut a social media reel from a collection of clips and images.
- Turn the assets of a generated script into a scene-based video.
For exact, manual control over up to four video layers, use Composite. To join clips one after another, use Combine Videos.
Quick start
Add the node
Press Tab on the canvas and choose Video › Cut & Assemble › Compose Video.
Connect the media
Wire images, clips and audio into the Media input. The input takes many connections.
Order the media
Open the settings panel and drag the items in the Media Order list. The items at the top get more priority in the composition.
Write the composition prompt
Describe the video in the Composition Prompt: the style, the timing, the transitions and the layering. Refer to a connected item by its name with @.
Run it and render
Click Run on the node to create the plan. Then wire the Composition output into a Render Video node and run that node to get the video.
Inputs
| Input | Accepts | What it does |
|---|---|---|
| Media | Image, video and audio nodes | The media to compose. Connect as many items as you need. |
The output, Composition, is the composition plan. Connect it to a Render Video node to produce the video file.
Settings
| Setting | What it does |
|---|---|
| Media Order | The connected media. Drag the items to set their priority: the model features the items at the top more. |
| AI Model | The language model that plans the composition. The default is Claude Sonnet 5.5. |
| Reasoning Effort | Shown for models that can reason. Auto (model default) is the default. The levels vary per model. A level has no price of its own: a higher level usually writes more tokens, and those tokens are charged. |
| Advanced mode | Available on Gemini and Claude models. Runs the model directly with the model maker, which unlocks Temperature, Max Tokens and the full range of reasoning depth. The call is then charged at the rate of the maker's own service. The node's estimate updates as soon as you switch it on. On a Claude model, choosing any effort level runs the model the same way, at the same rate. On other models, the switch is visible but disabled, with the reason shown. |
| Composition Prompt | The description of the video you want. Mention connected items by name with @. |
Under Advanced Settings, you set the canvas:
| Setting | What it does |
|---|---|
| Aspect Ratio | 16:9 (the default), 9:16, 1:1 or 4:5. |
| FPS | 24 fps (Film), 30 fps (Standard), the default, or 60 fps (Smooth). |
| Duration (seconds) | The total length of the composition, from 1 to 300 seconds. The default is 30 seconds. |
| Background Color | The color of empty areas. The default is black, #000000. |
After a run, the settings panel shows a preview of the composition.
Credits
The composition plan is one call to the AI model. The call is charged for the tokens it actually used, at the selected model's rate.
- Before the run, the Run button shows an estimate (
≈and a number) of a typical plan on the selected model. - A hold comes first. When the run starts, Nodaro holds enough credits for the longest answer the model may write. After the call, the run is charged what it used.
- Advanced mode, and any effort level on a Claude model, run the call on the maker's own service, at that service's rate.
- Failures. A failure that the model's maker billed is charged for that usage. A failure that was not billed is refunded.
The full rules are in How LLM calls are charged. Rendering the plan with Render Video is priced separately.
Tips
- Name your media in the prompt. Refer to connected items with
@and their node name, so the model knows which item goes where. - Be specific about timing. Explicit cues work best, for example
show image A for 5 seconds, then crossfade to video B. - Match the length to the media. A long composition with few items gives a sparse timeline. Keep Duration (seconds) in proportion to what you connect.
- Choose the shape for the platform. Use
16:9for standard video,9:16for vertical mobile video and1:1for square social posts. - Keep the temperature low. In Advanced mode, keep Temperature at about 0.5 or lower. Compose Video asks the model for structured output, and higher values start to break the format.
Frequently asked questions
Does Compose Video output a finished video?
No. It outputs a composition plan. Connect its Composition output to a Render Video node and run that node to get the video file.
Which AI model does Compose Video use?
The default is Claude Sonnet 5.5. You can choose another language model under AI Model in the settings panel. The plan is charged at the rate of the model you choose.
How do I make the AI feature a specific image or clip?
Drag that item to the top of the Media Order list, and mention it by name in the Composition Prompt with @, for example @product-shot.
How much does Compose Video cost?
The plan is one call to the AI model, charged for the tokens it used at the model's rate. The Run button shows an estimate of a typical plan before you run. Render Video is priced separately.
What if the composition does not match what I wanted?
Refine the prompt with explicit timing, for example "show image A for 5 seconds, then crossfade to video B", and run the node again.
Related
Render Video
Composite
Slideshow
Generate Text - Any LLM
Combine Videos
Last updated on
Mask Composite
Place a foreground video over a background through a mask video, or pixelate or blur the masked area to hide faces. It runs on every edition without a key.
Split into Chunks
Cut a video or audio file into frame-exact parts between a minimum and a maximum length, or a video into one part per shot, and run the next node on each.