This LoRA aims to recreate the character "Moika" as she appears in the animated music video "MOIKAILOOP x GHOST DATA". The two different versions provided focus on different levels of detail.

• AMV-Accurate: Aims to recreate her appearance as seen in the style of the original animated music video. It was produced on a smaller dataset for the same reason, so some smaller details are lost (earrings shape, tiny hair clip, etc.). If your intent is to depict this character as nude, consider adding completely nude to your positive prompt and chest tattoo, chest logo to negative.

• AMV-Divergent: Aims to recreate her appearance, outfit, etc. but slightly deviates from the style depicted in the original music video. I found that this version struggles to separate the character from her outfit.

In my opinion AMV-Accurate version is better, mainly because I used manual tagging instead of relying on an autotagger. The AMV-Divergent version might have more adaptability, however.

Ultimately this is just another LoRA that I made while I am figuring stuff out. This is not a comprehensive model of Moika herself; if you would like that then I recommend Konan's version located here.