First attempts at creating a LoRA. Tested to work pretty well on anime style models. I find a weight of 0.4 to be my personal preference and a full weight of 1 will massively distort the image. I also prefer to use it in img2img vs originally in the txt2img with around 0.4 denoising strength.