Trained on about 20 images, mostly fan art with a few AI generated ones thrown in for the photographic look. Captioned with natural language.









Sergals! It struggles with realistic/photographic, probably due to a lack of good training images, but it can handle a lot of other styles.
Prompting tips:
It really likes Juggernaut XL. Other models always seem to think I'm asking for a wolf or a dragon. Try negative prompting "black nose" if they show up with a black wolf nose.
Generally responds well to both "blue sergal" and putting "blue fur" later in the prompt. I've been writing "a luxuriously fluffy sergal" and adding "plastic, toy, doll" to the negatives. I didn't test if it actually helped, but it made me feel better!
I didn't train it on regional differences, so "eastern sergal" isn't going to do anything unless your base model has something.
If the image saturation goes crazy, CFG Rescale can help.
Trained on about 20 images, mostly fan art with a few AI generated ones thrown in for the photographic look. Captioned with natural language.
1. 轉载模型僅供學習與交流分享,其版權及最終解释權归原作者。
2. 模型原作者如需認領模型,請通過官方渠道联系海藝AI工作人員進行認證。我們致力於保護每一位創作者的權益。 點擊去認領
