> 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/create-on-the-canvas.md).

# Create on the canvas

Upload images, videos and 3D files, and make new images, videos and style presets straight on the canvas, without building a scene first.

Upload your own files, or generate images, videos, style presets, 3D models and worlds, straight on the canvas. Not every job needs a 3D scene. Sometimes you need a quick concept image, a style to test, a video from a prompt, or a place to collect references before anything gets built.

## Upload

* **Upload media from local** in the bottom toolbar adds files to **Sources**.
* **Add to canvas from media library** picks something you or your team already uploaded.
* **Upload into group** on a media group adds files straight into that group.

Both upload buttons take images (jpeg, png, webp, avif, bmp), videos, and 3D models and splats (glb, gltf, fbx, obj, dae, ply, spz, sog). GIFs are not accepted. Images and videos become media cards. A 3D model or splat is imported and lands as a card, without being placed in a scene; importing needs a plan that includes it. PDFs go in on Home, when you start the project. See [Home](https://help.intangible.ai/canvas/home.md) for every upload route.

Each uploaded card keeps its file name, extension included, for example `red-car.png`.

## Make something new

Click **Generate** in the bottom toolbar, or **+** on a media group, and pick what to make:

| Pick                            | What you set                                                                                    | What lands          |
| ------------------------------- | ----------------------------------------------------------------------------------------------- | ------------------- |
| **Image**                       | Prompt, model, aspect ratio, resolution, and reference images if you want them                  | An image card       |
| **Video**                       | Prompt, duration, aspect ratio, audio, model                                                    | A video card        |
| **Style preset**                | A name, an instruction and reference images                                                     | A style preset card |
| **3D** (bottom toolbar only)    | The same settings as [Generate 3D Asset](https://help.intangible.ai/build/generate-3d-asset.md) | A 3D model card     |
| **World** (bottom toolbar only) | The same settings as [Generate World](https://help.intangible.ai/build/generate-world.md)       | A world card        |

An image or video card appears straight away, so you can keep working. While it renders, it shows an animated cube of dots and a label with its name and the time since it started, such as `shot 04 · 12s`. When the job finishes, the picture forms out of the dots and the card fills in. Started from a media group, the result lands in that group; started from the bottom toolbar, it lands in **Sources**.

If an image or video fails, its card reads **Didn't finish**. Click **Retry** on the card to send the same request again, or select it and click **Delete**. See [Groups and cards](https://help.intangible.ai/canvas/groups-and-cards.md#while-a-card-renders).

A media group's **+** offers **Image**, **Video** and **Style preset**. To make a 3D model or world, use **Generate** in the bottom toolbar.

![Left, a media group's + menu with Image, Video and Style preset above the group toolbar. Right, the bottom toolbar's Generate menu, which adds 3D and World](https://1179478100-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FSWrjIqlgVAblx6vHGJiM%2Fuploads%2Fgit-blob-c067ffb87704f17418405ecc0446a808e0f0e507%2Fcanvas-group-generate-menu-01.png?alt=media)

**3D** and **World** open the same panels as [Generate 3D Asset](https://help.intangible.ai/build/generate-3d-asset.md) and [Generate World](https://help.intangible.ai/build/generate-world.md) in Build. The card can't be selected until the job finishes, and if the job fails, the card goes away.

### Make an image from reference images

An image can start from pictures you already have, such as a product shot and a location photo. The model builds a new image from them and your prompt.

1. Click **Generate** > **Image**. A panel opens headed **Generate · image**, **New image**, with the prompt and a row of chips under it: reference images, model, resolution, aspect ratio and **Generate**.
2. Click the image button at the left of the chip row (**Add reference images**). The **Reference images** section opens above the prompt; click the button again to hide it.
3. Add images with **Select from canvas**, **Media library** or **Upload**. With Select from canvas, click the image cards you want, then click the tick to add them.
4. Write a prompt and click **Generate**. A prompt is required.

Each model takes a set number of images, and the panel says how many. With more than the limit, Generate stays off until you remove some.

The aspect ratio chip sits after the resolution, just before **Generate**, and shows the current ratio, such as 16:9. Without reference images it starts on 1:1. With references attached it offers **Auto (match first reference)**, which measures the first reference image and sends the model's closest ratio, so a wide location photo comes back wide. The list is the model's own, and it can grow once references are attached, because the model's edit version runs then (Nano Banana goes from 5 ratios to 10). If you switch to a model that lacks your ratio, the panel picks one it has, and your ratio comes back when you switch to a model that offers it.

The options a video offers (durations, aspect ratios) depend on the model you pick, and change when you switch models. Generating spends credits. See [Credits and tokens](https://help.intangible.ai/learn-more/concepts/credits-and-tokens.md).

## Style presets on the canvas

A style preset shows on the canvas when this project created it, placed it in a group, or used it for a render. Select one for **Edit** and **Delete**. Editing a preset changes it everywhere, because presets are shared across the team.

{% hint style="warning" %}
Deleting a style preset asks **Delete from library?** It deletes the preset itself, and it is removed for every project that uses it. This can't be undone.
{% endhint %}

The rest of the team's presets stay in the style picker. See [Style presets](https://help.intangible.ai/visualize/style-presets.md).

## Then what

From there:

* [Edit media](https://help.intangible.ai/canvas/edit-media.md) changes part of a generated image or video.
* [Chat with Intangible](https://help.intangible.ai/chat-agents-intangible/chat-with-intangible.md) to use an image as a reference.
* [Asset packs](https://help.intangible.ai/canvas/asset-packs.md) hand it off.

## Related

* [Groups and cards](https://help.intangible.ai/canvas/groups-and-cards.md)
* [Generate 3D Asset](https://help.intangible.ai/build/generate-3d-asset.md) and [Generate World](https://help.intangible.ai/build/generate-world.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/create-on-the-canvas.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.
