> For the complete documentation index, see [llms.txt](https://luw-ai.gitbook.io/luw.ai-app-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://luw-ai.gitbook.io/luw.ai-app-docs/working-with-images/generate-an-image.md).

# Generate an image to start from

Make a starting picture from a written description inside a tool's upload area, for when you don't have a photo yet.

This page explains **Generate Image with AI** in the upload area, which makes a picture from your description and uses it as the tool's photo.

## When to use it

* **You don't have a photo yet.** Describe the room, building or object, generate a picture, and work on that.
* **You want a 3D model from words.** 3DGen needs a photo to start from. Generate a picture of the object here first, then make the model. See [3DGen AI](/luw.ai-app-docs/3d-and-video/3dgen-ai.md).
* **You want to try a tool with an idea** before you have real pictures of the space.

## Make a starting image

1. In the empty upload area, click **Generate Image with AI** (under "OR"). The upload area turns into a small panel.
2. Describe the picture you want in the **Prompt** box. To get help writing it, click the ArchiGPT button (the sparkle in the corner of the box). It asks you to confirm ("1 token = \~3000 words"), then opens an ArchiGPT chat, and the prompt it writes goes into the box.
3. Turn on **Symphony** if you want that engine. It's off to start with.
4. Click **Generate Image with AI**.
5. Wait for the picture. It can take up to a minute or two, so leave the panel open until it appears.

<figure><img src="https://1930792265-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F4yESzclcaZeSiIDZ8eIa%2Fuploads%2Fgit-blob-a9e4fe9500c77f84ef4a2ea126d5c72989eb7b5e%2Fgenerate-image-panel.webp?alt=media" alt="The Generate Image with AI panel: an empty Prompt box with the ArchiGPT sparkle button, a Symphony switch turned off, a Generate Image with AI button, a Close button and the line 1 token / generation"><figcaption><p>Prompt, Symphony switch and the credit line, all inside the upload box</p></figcaption></figure>

If you aren't signed in, clicking the prompt box takes you to the sign-in screen.

## Keep it or try again

When the picture is ready, it appears next to three buttons:

| Button            | What it does                                                                                                                      |
| ----------------- | --------------------------------------------------------------------------------------------------------------------------------- |
| **Save**          | Uses the picture as your photo. It's uploaded like any other photo, and the form moves on to the next step.                       |
| **Cancel & Undo** | Asks you to confirm, then throws the picture away and takes you back to the prompt so you can try again.                          |
| **Close**         | Closes the panel and returns to the empty upload area. If a picture is showing or still being made, it asks you to confirm first. |

<figure><img src="https://1930792265-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F4yESzclcaZeSiIDZ8eIa%2Fuploads%2Fgit-blob-3bdb4063b6f5da50daee3c512834c4d6911dec73%2Fgenerate-image-result.webp?alt=media" alt="A generated flat-lay of oak boards, linen swatches, terracotta tiles and a brass lamp, with Save, Cancel and Undo, and Close buttons beside it"><figcaption><p>Save keeps the picture as your photo; Cancel &#x26; Undo lets you try another prompt</p></figcaption></figure>

## Credits

The credit line under the buttons reads "1 token / generation". With **Symphony** on, each picture uses 2 credits. Every time you click **Generate Image with AI**, a new picture is made and charged.

## Tips

* Describe the kind of picture the tool works with: a room for Interior, a house front for Exterior, a single object for 3DGen.
* Wait for the result rather than clicking Generate again, so you don't pay for a second picture.
* If you only want a picture and not a redesign, Fluw makes images from text as finished results. See [Fluw](/luw.ai-app-docs/render-and-create/fluw.md).

## Related

* [Uploading images](/luw.ai-app-docs/working-with-images/uploading-images.md)
* [3DGen AI](/luw.ai-app-docs/3d-and-video/3dgen-ai.md)
* [Credits and engines](/luw.ai-app-docs/getting-started/credits-and-engines.md)