> If the AI can't recreate it exactly, it can recreate a likeness that is compelling enough that the average person would think it was the same
That's the opposite goal of this image model. Sure you might find other types of research models which are meant to do that but that's not stablediffusion and the likes.