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?