> It would involve a chain of attribution per fraction of a pixel through all it's past iterations. Which wouldn't be sufficient since, as stated many times be

> It would involve a chain of attribution per fraction of a pixel through all it's past iterations. Which wouldn't be sufficient since, as stated many times before, the diffusion process most text-to-image AIs use is not collaging.