Skip to main content
Image (3) 1

What is Image Prompt?

Think of an Image Prompt as the “visual shorthand” for your creative ideas. Instead of relying solely on words, you provide a starting image; a photo, a sketch, or even a previous generation to act as a structural anchor. You add an image from your uploads, your creations, or a stock library, and the AI uses it as a foundational layer of inspiration. It’s like a conversation where your image sets the scene, and your text prompt provides the direction.

How to use Image Prompt

1

Open the asset picker

In Image Studio, click + at the left of the prompt bar. The asset picker opens.
Prompt bar with the + button, the Describe the image you imagine prompt box, mode tabs, model button, Add references, and Generate
2

Choose your reference image

Pick an image from one of the picker’s tabs, then click Select. The button shows how many images you’ve picked, for example Select (1).
  • Uploaded assets — images you’ve uploaded before. Click Upload media to add a new file from your device. Supported formats include JPEG, PNG, and WEBP.
  • My creations — images you’ve generated in ImagineArt.
  • Unsplash and Pexels — free stock photo libraries. Search with Search assets or browse categories such as Abstract, Nature, Architecture, and Portrait.
  • Pinterest — images and videos from your Pinterest boards and saved pins.
Asset picker with Uploaded assets, My creations, Unsplash, Pexels, and Pinterest tabs, a Search assets field, category filters, and a grid of Pexels photos
The first time you open Pinterest, click Connect Pinterest to link your account.
Pinterest tab in the asset picker with a Connect Pinterest button
For best results, use a clear, well-lit image that reflects the composition or style you want the AI to draw inspiration from.
3

Write your text prompt and tag the image

Your image appears as a thumbnail above the prompt box. Adding an image switches the prompt bar to the Edit tab.Describe what you want, and tag the image where it belongs in the prompt: type @, then pick it under Images. Images are named in the order you add them (Image1, Image2, and so on). Tagging makes it clear which image you mean when you add more than one.
Prompt reading "A rainy evening on Image1 street, cinematic lighting" with the Image1 tag chip and the image thumbnail above the prompt
Example combinations:
  • Reference image of a misty forest + prompt: "A lone cabin with warm glowing windows, winter evening, oil painting style"
  • Reference image of a sports car + prompt: "Same car on a coastal highway at sunset, cinematic photography, motion blur"
  • Reference image of a character sketch + prompt: "Full color illustration, fantasy armor, dramatic lighting"
4

Select your model and settings

Click the model button in the prompt bar to open Settings and choose a model and aspect ratio. Not happy with the first result? That’s completely normal — and part of the fun! You can adjust your text prompt, swap out your reference image, or tweak your settings and generate again. Each iteration builds on your feedback, helping you get closer to your ideal image with every attempt.
Settings dialog model list with Search models, Providers and Resolutions filters, and Featured models
5

Click Generate

When you’re happy with your image prompt and text, click Generate to generate your image. The AI will combine your visual reference and text instructions to produce a unique result. Depending on the selected model and settings, generation typically takes just a few seconds.
Generate button with its hover tooltip showing the generation's credit cost

Support

Image prompt is supported by all available image generation models. Some examples are below; see Create Image for the full list.
The model you choose affects how strongly the reference image influences the output. Experiment with different models to find the balance between reference fidelity and creative freedom that works best for your project.