Skip to main content
What makes a good prompt:
  • Subject — What is in the image. “A soda can”, “a futuristic cityscape.”
  • Style — How it should look. “Minimalist”, “photorealistic”, “watercolor.”
  • Details — Colors, lighting, composition, mood.
  • Context — Where it will be used. “For an Instagram post”, “as a website hero.”
This guide walks you through generating images from a text description. You describe what you want, choose an aspect ratio, and the agent creates one or more images on the canvas.
1

Select Image as your artifact type

On the home dashboard, click the artifact picker and select Image.
Home dashboard showing the artifact picker with the Image option highlighted
2

Choose your aspect ratio

Select an aspect ratio from the options below the artifact picker:
RatioDescriptionBest for
21:9Ultra-wideBanners, cinematic scenes
16:9Widescreen (default)Presentations, thumbnails
4:3StandardBlog images, general use
1:1SquareSocial media posts
3:4PortraitPosters, product shots
9:16Tall portraitStories, vertical content
9:21Ultra-tallVertical banners, wallpapers
Aspect ratio dropdown with options from 21:9 to 9:21
3

Describe your image

Type a detailed description in the input bar. The more specific you are, the better the result.

Product mockup with branded packaging.

Minimalist branding design.

Cinematic AI-generated photography.

Social media infographic.

You can also attach reference images using the + button. The agent uses them for style reference, composition guidance, or content to incorporate.
4

Generate multiple images (optional)

If you want more than one image, say so in your prompt. For example, “Generate 3 variations of a logo for a coffee brand.” The agent generates all images in parallel and places each one on the canvas.
5

Review on the canvas

Generated images appear on the canvas in the side panel. Click any image to view it at full size.
Canvas side panel showing three generated image variations with one enlarged in preview mode
If an image is not what you wanted, you have several options:
  • Ask the agent to refine it - Describe what to change in the chat.
  • Generate again - Ask for a new version with an updated prompt.
  • Edit directly on the canvas - Select the image and use canvas tools for targeted adjustments.

​
Tips for better results

“A clean product photo of a white sneaker on a wooden table, soft natural light” works better than “A sneaker photo, no busy background, not too dark.”
Put the visual style near the beginning. “Photorealistic close-up of a coffee cup” gives clearer direction than “A coffee cup that looks photorealistic.”
If you have a specific look in mind, attach a reference image. The agent can match styles and compositions from references much more accurately than from text alone.
Ask for 2–3 variations in a single prompt to compare directions. “Generate 3 variations of a tech startup logo, minimal style, blue tones.”

​
Download your image

Click any image on the canvas, then click the Download button to save it to your device. Images download as PNG files.
If you need a higher resolution version before downloading, upscale the image first. See Upscale an Image.

​
Next steps

Remove Background

Strip the background from your generated image with one click.

Replace Background

Swap the background for a new AI-generated one.

Edit Text in Images

Detect and change the text inside any image.

Mark & Edit

Draw on specific areas for precision editing.