> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cleve.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Image generation

> Ask Cleve to create or edit images in chat and add them to your docs

Cleve can create images from a description and edit existing ones, right in chat. Use it for doc headers, illustrations, cover images, mood boards, or a quick visual for a slide.

> [!NOTE]
>
> Image generation is available on paid plans. See the [pricing page](https://www.cleve.ai/pricing).

## Create an image

Ask for it in chat, in your own words:

* *"Create a header image for this doc: a calm watercolour of a harbour at dawn."*
* *"Make a flat illustration of a team planning on a whiteboard, in blue and orange."*
* *"Design a simple poster that says 'Launch week' in bold letters."*

While it works, the reply shows **Create Image** and a placeholder in the shape of the image. An image usually takes between 10 seconds and a minute. When it's ready, it appears in the reply.

Cleve writes a full description for the image model from your request, so you don't need to write a detailed prompt yourself. It helps to mention:

* **Subject and style**, for example photo, watercolour, flat illustration, or 3D
* **Colours and mood**
* **Any text** that must appear, in quotes
* **Where it's going**, such as a doc header or a phone wallpaper, so Cleve picks a good shape

## Edit an image

Cleve can change an image instead of starting over. Say what to change and what to keep:

* **The image it just made:** *"Make the sky darker and remove the boat."*
* **An image in a doc:** *"In my Launch plan doc, change the header image to evening light."*
* **A photo you attach:** attach an image to your message, then ask, for example *"Turn this into a pencil sketch."* Cleve edits the latest image you attached in that chat.

Cleve can only edit images stored in Cleve: ones in your docs, ones it created, or ones you attach. It can't edit an image from a web link.

While it works, the placeholder says **Editing image…**.

## Add an image to a doc

If you're working in a doc and want the image there, say so: *"Add it to the top of this doc."* Cleve inserts it where it fits.

Inserting into a doc is an edit, so it needs **Edit mode**. In **Chat mode**, Cleve can still create images in chat but won't add them to your docs. See [Chat mode and Edit mode](https://docs.cleve.ai/features/ai-chat.md#chat-mode-and-edit-mode).

## Shape and quality

### Aspect ratio

Cleve picks the shape to suit where the image will go. You can also ask for one:

| Ratio | Good for |
| - | - |
| **1:1** (default) | Square images, avatars, social posts |
| **16:9** | Doc headers, banners, slides |
| **9:16** | Phone screens and stories |
| **4:3** | Classic photo and slide shape |
| **3:4** | Portrait images |
| **21:9** | Extra-wide banners |

### Quality

| Quality | When to use it |
| - | - |
| **Standard** (default) | Most images. Faster, and uses less of your usage. |
| **High** | Sharper detail, and better at legible text such as posters, slides, infographics, and signs. Takes longer and uses more of your usage. |

Cleve uses standard quality unless you ask for high quality or the image needs readable text.

> [!TIP]
>
> For charts and diagrams that need exact numbers or labels, ask for a Mermaid diagram or an interactive block instead. They stay accurate and editable. See [Blocks](https://docs.cleve.ai/features/blocks.md) and [Interactive blocks](https://docs.cleve.ai/features/interactive-blocks.md).

## Download an image

Hover over the image in chat and click the download button in its top-right corner. Click the image itself to open it full size in a new tab.

## Limits

* **Up to 4 images per reply.** If you want more, ask again in a new message.
* **Usage:** each image uses some of your plan's usage, and high quality uses more than standard. If you don't have enough usage left, Cleve tells you, and you can try again once your limit resets. See [How usage works](https://docs.cleve.ai/pricing/usage.md).

## Content rules

Whatever the request, Cleve won't:

* Create sexual or nude content
* Create or edit sexualised, violent, or demeaning images of real people. This includes photos you attach and any image of a recognisable person.

When a request breaks these rules, Cleve declines in its reply instead of creating the image, and doesn't try a watered-down version. The image model can also decline a request on its own. In that case Cleve tells you, and you can reword the request and try again.

## Related

> **[Chat with Cleve](https://docs.cleve.ai/features/ai-chat.md)**
>
> Attach images and give Cleve context.

> **[Blocks](https://docs.cleve.ai/features/blocks.md)**
>
> Images, diagrams, and more in your docs.

> **[How usage works](https://docs.cleve.ai/pricing/usage.md)**
>
> What uses your plan.

> **[Plans](https://docs.cleve.ai/pricing/overview.md)**
>
> Which plans include images.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.