The DLSS 4 Transformer model, preset K, looks over-sharpened and introduces a bunch of artifacts. It looks bad. It also runs worse than the CNN model, which is blurrier but still looks better. The transformer model has received almost nothing but universal praise despite this. It's like how people are unironically asking for and suggesting frame generation. Can they just not see how awful and unnatural it looks? Are their brains wired differently?