Skip to main content
You can generate images directly on the canvas without going through the chat. Click a button, type a prompt, pick a model, and the image appears on your workspace. This is faster than asking the agent and gives you more control over model selection and placement.
Generating directly on the canvas gives you more control over which model to use and where the image is placed. It also uses fewer credits than sending a prompt to the agent for simple generation tasks.

How to generate an image on the canvas

1

Click Generate Image

In the bottom toolbar, click the Generate Image button. A placeholder appears on the canvas at the next available grid position.
Canvas bottom toolbar with the Generate Image button highlighted
2

Enter your prompt

A prompt box appears below the placeholder. Type a description of the image you want.

Product photography on a clean background.

Illustration for a blog post.

3

Choose a model and aspect ratio

Below the prompt field, select your preferred image model and aspect ratio.
OptionWhat to choose
ModelPick the model that fits your needs. Each has different strengths (color accuracy, speed, 4K output, etc.)
Aspect ratio21:9, 16:9, 4:3, 1:1, 3:4, 9:16, or 9:21
Canvas prompt box showing a text field, model dropdown, and aspect ratio selector below a generating placeholder
4

Submit and wait

Click the submit button. The placeholder shows a generating status while the AI creates the image. Once complete, the image replaces the placeholder on the canvas.

Edit the generated image

After the image appears, select it and use any of the canvas editing tools:
ToolWhat it does
Remove BGStrip the background
Text EditDetect and change text in the image
Mark EditDraw on regions and describe targeted changes
UpscaleIncrease resolution 1x, 2x, 3x, or 4x
Edit ImageOpen a new prompt box to modify the image with AI
Create VideoAnimate this image into a video clip
All edits create a new item. Your original is preserved.

What you cannot do

  • You cannot generate multiple images at once from the canvas. Generate them one at a time. For batch generation, use the agent in the chat.
  • You cannot set a specific pixel resolution. The model determines the output size based on your aspect ratio.

Next steps