Sample images were generated with clip skip 2, DPM++ 2S a Karras. Weight 0.8 is usually best.
Anime models typically work well with cowgirl position, but SD in general is bad with hands and hand poses. For my first ever public LoRA, I decided to fix this problem in the case of one particularly based pose. Don't think about what this says about me.
The model is fairly flexible and, while there is the usual element of lottery, it is fairly consistent in my experience. It does need good prompting, and the specific tags you need to use will probably vary by base model. The provided tags are a good starting point.
Feedback would be greatly appreciated since I am new :)