>>106255529
Actually, you are very wrong. This whole "fried" thing is ridiculous. You are wrongly expecting every image to be sharp, because that is what every AI image generically looks like. But that's not what real images scattered throughout the web look like. That is only what nitpicked stock images look like. It's no surprise that models that can only do clean type of images are slopped. That's because they do not possess a wealth of images in their dataset. In other words, they are censored, and therefore slop. A LoRA can be trained for Chroma to create any kind of "clean" images you'd like. I like and prefer my "rough" images.