The generative fill is awesome but its lacking one major thing. Noise matching.
everything produced with the ai just looks blurred or lacking the natural noise of the photo, and matching it manually isnt perfect.
so noise matching would be great!
another awesome idea i have: since the render is 1024x1024 and above that its scales up the results causing it to be lower quality i thought maybe add a checkbox for "High Quality" renders which renders the area with 1024x1024 patches automatically and that way we end up with native resulotion quality in exchange for time of process.
also maybe add a refine feature so the AI will render the same thing just in higher quality and not something different