To use an image as a reference in Leonardo AI, select Image Reference, upload your image, and write a prompt describing the change you want. The uploaded image gives Leonardo visual information to work with alongside your text.
For this demonstration, I use an Egyptian cat image and ask Leonardo to turn the cat into a Greek god. First, there’s an image-selection warning to keep in mind.
Choose an image Leonardo AI will accept
In my test, Leonardo didn’t accept an image associated with a popular movie. I mention Minions as an example of the kind of reference that can cause a problem, even when you’ve looked for the original source or credit.
That’s an observation from this test, not a complete explanation of Leonardo’s content rules. For the walkthrough, I switch to an Egyptian cat image I have available.
The interface shown allows up to six reference images, but this example only needs one.
Upload your reference image
Start in Leonardo’s image-generation screen. The upload begins with the small image icon beside the prompt area, which Leonardo also describes in its official image-upload guide.
- Select the small image icon.
- Choose Image Reference.
- Select Upload an Image.
- Pick the image you want to use.
Once the Egyptian cat image is attached, it’s the reference for the new generation. The next part is telling Leonardo what to change.
Describe the transformation you want
Start with a short prompt
The prompt in this example is direct:
Turn this cat into a Greek god.
You don’t need a long description for this demonstration. The reference supplies the cat’s existing appearance, while the prompt names the transformation.
This is a reference-led workflow, one of the approaches covered in our Leonardo AI review. Here, the task stays focused on changing the uploaded cat’s appearance.
Review the prompt and image-shape options
The demonstration also shows an Auto setting, additional options, and a prompt-improvement feature. I try the prompt-improvement option to expand the wording of the original request.
You can also change the image dimensions before generation. The options shown include 16:9, 4:3, and 9:16, so the output doesn’t have to keep the same shape throughout your experiments.
These adjustments sit alongside the two main inputs: the reference image and the written request.
Generate the image and compare it with the reference
After setting up the reference and prompt, generate the new image. In this example, the requested change is a Greek-god-style version of the Egyptian cat.
I then open the result and return to the original reference for comparison. The close-up review focuses on the cat’s details and the background, rather than looking only at the new overall appearance.
That comparison is the central part of the demonstration: seeing what Leonardo changed in response to the prompt and examining which parts of the reference remain recognizable.
Try a reference-led transformation
The Egyptian cat example uses one reference image and a short prompt to request a different appearance. Uploading the image supplies the visual starting point; the text tells Leonardo the direction to take.
You can try the same workflow through Leonardo AI. Share how your image-reference test turns out in the comments.
For writing projects alongside your visuals, our Free AI Tools offer text-focused options.
