The paper vtuber with the cool design. The Lora will work on Illustrious and NoobAI, and all of their derivatives.
The hardest dataset I have gathered yet, with only 3/27 images on danbooru. This means there was a lot of manual tagging involved.
The Lora implicitely knows the horns, and occasionally gets her skin pattern and hair, however the tag "white skin" is basically required for it to work correctly.