Initial release.
Trained with 50 images at 10 repeats and 10 epochs.
Base model is NAI.
Training resolution is set to 768.


This is my first attempt at creating a LoRA. As the title and images suggest, this is a LoRA of Gavial from Arknights. I made this for personal use, but I figured other people might find it useful.
The base model is NAI, but it should work with most anime models like AnythingV3 and AOM2. Just remember to set CLIP skip to 2. I recommend setting the weight to between 0.6-0.8. 0.5 and below tends to be too weak, and 1.0 is probably too strong. Of course, depending on your prompts and model, your mileage may vary.
When unspecified, characters tend to be generated with swimsuit-like clothing, but other outfits should be possible if prompted appropriately.
Initial release.
Trained with 50 images at 10 repeats and 10 epochs.
Base model is NAI.
Training resolution is set to 768.
1. 재게시된 모델의 권리는 원 제작자에게 있습니다.
2. 모델 원작자가 모델을 인증받으려면 공식 채널을 통해 SeaArt.AI 직원에게 문의하세요. 저희는 모든 창작자의 권리를 보호하기 위해 노력합니다. 인증하러 이동
