It generates a few similar images based on the input image. You can choose how similar you would like generations to be between Low, Medium and High. Choosing similarity can be a bit tricky, a good idea is to start from Medium similarity. If generated images look good, perfect! On the other hand, if they feel too similar, then try Low similarity. Or High similarity if the feel too varied.

It works really on top of your own digital artwork as well as generations made from Creative model. It enhances Creative model generations quite a lot, like fixing some AI artifacts.

Prompt works differently in Similar model where it should describe your input image rather than what you want to generate.

Let's look at some examples.

Input Image has been generated by using Creative Model on top of Drafts using prompt I am lost

Face is a bit distorted, passing through Similar model should fix it. We will run Similar model on it with different similarity settings.

Similar model with Low similarity

Similar model with Medium similarity

Similar model with High similarity