This lora model is my little experiment, playing around with a new way to gather dataset using cutscenes from Resident Evil 2, 3, and 4.
The result image can capture characters face but not really good due to the face quality of the cutscene.
It's a work in progress, not exactly fantastic yet, but I thought, "Hey, why not share it anyway?" Who knows? Maybe I'll cook up some improvements in the next version—if time and ideas decide to play nice.