@CursedSilicon I don't think anybody could afford to do "AI upscaling" on a commercial data-center cloud account if it actually used an LLM.

It's conventional ML, marketed as "AI"? Don't throw out the baby with the bath water... the ML algorithms are generally efficient and powerful and not as deeply ethically troubling when compared to LLM's.

The proof is in the pudding: Does it look like generative slop? Does this person have a good reputation and critical eye? Do you like the results?