I'm curious as I have not idea how you do this, but do you prompt a GenAI model with an image to generate an output image? What happens if the image is not 1:1 accurate?

I guess they’ll text me or stop using the service. With food is much better.

Even in the demo image in the hero section you can notice that the "AI generated" costume is not the same as the "real image" one.

Yes, it derails a bit when it has to reproduce creative patterns