If someone has photos of a person, those photos could help generative AI to fix a bad photo. Also it would be nice when using (object select) on a blurred picture you could set a point where the blur ends. That point could be used to indicate where to blend to reduce or get rid of the blur.