Anyone who has worked with AI image‑generation technology is more or less aware of some paid upscaling models. Today I saw a well‑known model/tool release a new version. Coincidentally, I have also been training a new high‑definition model lately, so on a whim I decided to run a comparison test with it. The result was somewhat unexpected…

Static image comparison

Judging from the comparative test results, the paid model seems to perform better in background details and blurring effects. However, when it comes to the detailed rendering of characters and skin texture, I prefer my LoRA. What do you think? Feel free to leave a comment and share your thoughts.

Oh, right, I checked out the pricing of paid models, and suddenly I'm really glad LoRA exists!

Long live LoRA!