> For the complete documentation index, see [llms.txt](https://help.intangible.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.intangible.ai/canvas/keyframes-to-video.md).

# Keyframes to video

Turn a keyframe on the canvas into a video shot, by asking in the chat or through Edit in Visualize, then skim and play the result.

A keyframe shows what a shot looks like. A video shows how it moves. Because every shot on the canvas comes from a 3D scene, the motion you set up there (a camera move, an animated object, a simulation) can drive the video, so the clip moves the way the scene does instead of the way a model guesses it should.

## Ways to make a video

| Start from                | Route                                                                                | What you get                                                 |
| ------------------------- | ------------------------------------------------------------------------------------ | ------------------------------------------------------------ |
| An image render           | **Edit in Visualize** in the render's toolbar, then make the video in Visualize mode | A video from the keyframe, with every video setting in reach |
| An image render or a shot | Ask in the chat panel                                                                | A video the agents set up and submit                         |
| A prompt only             | **Generate** > **Video** in the canvas's bottom toolbar                              | Text-to-video, with no keyframe or scene behind it           |

A shot card's **Visualize** button opens a panel that makes images only. Video that uses the 3D scene as its motion reference comes from the chat or from Visualize mode, not from the canvas panels.

## Make it in Visualize

1. **Select the image render** on the canvas.
2. **Click Edit in Visualize.** Visualize opens on that render's shot.
3. **Make the video** there. Pick **From scene animation** for a clip that follows the shot's camera move and animation, or **From Keyframes** to start from the render and let the model invent the motion. See [Generate video](https://help.intangible.ai/visualize/generate-video.md).

The video comes back to the canvas with the shot's other renders.

## Chat with Intangible for a video

1. **Name the keyframe.** Say which shot you mean the way the canvas names it.
2. **Ask for the video.** For example: "Render shot 01's keyframe as video, from its keyframe and scene."
3. **Agree to the cost.** The agents pick the mode and the video model themselves, weighing what the shot needs against credits. Before they start, they say in a line or two what you'll get and how many credits it costs, sometimes with one alternative such as a cheaper rough version. Nothing starts until you agree. They don't name the model unless you ask which one, and if you name a model yourself they use it when it fits. Models still in early access are only used when you name them.
4. **Find the video.** It lands in the shot's Visualize media list and as a new card in the **Visualizations** group.

A video animated from a render starts from that still and the model invents the motion; it does not follow the path the scene animates. When the shot is animated, the agents follow its motion on their own; to be sure, ask for the video to follow the scene.

In Visualize, **From Keyframes** means start and end frames, not the keyframes you set in Compose. It is the mode that invents the motion; **From scene animation** is the one that follows your scene.

You can ask for several at once: "Render shots 02, 03 and 05 as video, each from its keyframe and scene."

When the agents make a video, they write its video settings into the shot's Visualize panel, as the panel itself does, so you can open the shot in Visualize, see exactly what was used, change it and render again yourself. See [Generate video](https://help.intangible.ai/visualize/generate-video.md).

## Watch it back

A finished video card shows its poster frame at rest.

* **Skim:** hover the card and move the pointer across it. A line follows the pointer and the card's tab shows the time, for example `0:07 / 0:18`.
* **Play:** double-click the card to open the video full size over the canvas, and play it there. Cards themselves do not play.
* **Turn skimming off:** select the card and press **S**.

A video card that is still rendering shows an animated cube with its name and elapsed time, and skimming waits until the video has landed. A failed video card reads **Didn't finish**, with **Retry** when the request can be sent again. See [Groups and cards](https://help.intangible.ai/canvas/groups-and-cards.md#while-a-card-renders).

## Credits

Video costs more than an image and scales with length and model. When you render from the Visualize panel, a banner warns you before a large video renders. A video you ask the agents for skips that banner; they tell you its credits in the chat instead and wait for you to agree. See [Credits and tokens](https://help.intangible.ai/learn-more/concepts/credits-and-tokens.md).

## Related

* [Batch visualize](https://help.intangible.ai/canvas/batch-visualize.md) – make the keyframes first.
* [Edit media](https://help.intangible.ai/canvas/edit-media.md) – fix a frame or a clip before you deliver it.
* [Generate video](https://help.intangible.ai/visualize/generate-video.md) – full control over a video render in Visualize.
* [Deliver](https://help.intangible.ai/get-started/deliver.md)


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://help.intangible.ai/canvas/keyframes-to-video.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
