The AI Image Generation action now supports reference images. You can attach up to five images to guide the output alongside your prompt.

What’s included

• Up to five reference images per action: Add up to five images to a single AI Image Generation action. Each image is sent as visual context together with your prompt.
• Multiple ways to add images: Upload from your computer, choose from the Media Library, or provide an image URL.
• Dynamic reference URLs: The URL field supports custom values, so you can pull a reference image from a webhook payload, a contact field, or a previous step in the workflow.
• Ordering matters: Images are processed in the order you add them, so you can refer to “the first image”, “the second image”, and so on in your prompt.

How to use it

1. Open the AI Image Generation action in your workflow.
2. Write your prompt as usual.
3. Scroll to the Reference images section.
4. Choose a source (upload, Media Library, or URL).
5. Add up to five images.
6. Refer to each image by position in your prompt.
7. Save the action.

Example

If you want a lifestyle image without a photo shoot, you can upload reference images for the setting, the subject, and the logo. Then write a prompt that tells the model which reference image to use for each part of the scene.

Why this helps

Reference images make it easier to keep faces, settings, and logos consistent, and they reduce the guesswork that happens when you try to describe visual details with text alone. Using dynamic URLs can also help you generate personalized visuals at scale within a single workflow.