This lora is Dog shake, very simple.
The dataset 512*512 realistic image, it will perform poor in 2.5D and 2D arts.
