A custom SDXL checkpoint created by merging 10 different checkpoints. Trained specifically on a dataset of anime images and video frames focusing on hand ??????? and movement to improve hand generation accuracy and body part accuracy in general in anime/fantasy styles. That said we all know the limitations of AI on generating precise body parts with a common sense that is currently lacking this is my way trying to fix that and it begins here XD (gl to me ). Even if This is far from my concept of "good" I saw some decent progresses  with correct pre loaded image and smart prompt.