> 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/edit-media.md).

# Edit media

Fix part of an image or video on the canvas by drawing on it and writing a note, with optional reference images, without leaving the canvas.

Most renders are nearly right. Edit fixes the part that is not: add something that is missing, recolour one thing, or remove an extra object. You show the model where with a drawing and tell it what with a note.

## What an edit can do

* **Swap a character or object.** Mark it, attach a reference image of what should be there, and say so in the note, for example "replace this person with the woman in the reference".
* **Restyle or relight.** Describe the change in the prompt ("make it dusk, warm low sun from the left"), or attach a reference image with the look you want.
* **Fix a detail.** Add, remove or recolour one part of the frame.

Edit does not crop an image, trim a video or stitch clips together. Download the file and finish it in the image or video editor you already use; to join shots in order, export the scene or sequence ([Export Video](https://help.intangible.ai/publish/export-video.md)).

## Edit an image you uploaded

Every finished image on the canvas has **Edit image**: renders, images you uploaded, reference sheets and images the agents made. The first time you open Edit on an upload or a reference sheet it takes a moment, while Intangible makes a working copy; the original in your media library stays as it was. Uploaded videos have no Edit video button.

You can also ask the agents in the chat panel: name the card and say what to change, for example "Edit docs-test.png: make the car blue, keep everything else." The agents start the edit, and the edited copy lands in the shot's Visualize media list next to the original, which stays as it was. The edit uses credits, like any other image edit.

## Edit an image

1. **Select a finished image card** on the canvas.
2. **Click Edit image** in its toolbar. The editor opens with the image and every edit control in one place.
3. **Mark the area.** Pick a tool from the toolbar: **Select**, **Draw**, **Rectangle**, **Ellipse** or **Arrow**. Draw has four brush sizes, which appear next to it while it is selected. Pick a colour; one instruction can use as many marks and colours as it needs.
4. **Write the instruction** in the note that appears, for example "add a dust plume around the airbags". **Esc** saves the note and closes it; press **Esc** again to close the editor.
5. **Add references (optional).** A note can cite up to three reference images. Pick them from your media or the team's media, or upload one.
6. **Adjust and generate.** Change the main prompt, model or quality if you need to, then click **Generate**.

The edited image lands on the canvas as a new card, keeps the source's scene and shot, and becomes the selected card. The source stays as it was, so you can compare the two or edit again.

## Edit a video

Select a finished video card and click **Edit video**. A finished video open in Visualize has the same **Edit video** button. The editor is the same, with the clip and its playhead added:

1. Scrub to the moment you want to change.
2. Draw over it and write the instruction in the note. Notes on a video are tied to a time.
3. Adjust the prompt, model and quality, then click **Generate**.

The edited video lands as a new card and shows as generating until the job finishes. Video edits cost by duration.

## Swap in an edited reference sheet

An edit to a reference sheet lands as a new image card. The subject keeps the sheet it had until you swap the edit in. Ask in the chat, for example "I've updated the character sheets, can you reattach them?", and the agents attach the edited image in place of the old sheet. To make that possible, Intangible copies the image into your media library at no credit cost; the copy shows up as a card in **Sources**, named after its file.

## Come back to an edit

Closing the editor keeps your markup as a draft. Once an image or video has an edit, a pencil sits in the card's lower-left corner. Click it to reopen the editor with your marks and notes as you left them; the card does not need to be selected first. **Edit** on the card does the same.

## Chat with Intangible instead

For a change you can describe without pointing, ask for it in the chat panel and name the card, for example "make shot 02's keyframe dusk". Use Edit when the change belongs to one part of the frame.

## Related

* [Edit images](https://help.intangible.ai/visualize/edit-images.md) – the same editor in Visualize.
* [Keyframes to video](https://help.intangible.ai/canvas/keyframes-to-video.md)
* [Create on the canvas](https://help.intangible.ai/canvas/create-on-the-canvas.md)
* [Asset packs](https://help.intangible.ai/canvas/asset-packs.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/edit-media.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.
