This model adds fuzz and static to images, particularly faces, darkens backgrounds, and can give an image something like canvas grain. The number of steps you use to generate can make a big difference.