New dataset, suggested weight is ~0.4 or heads may disappear. close-up and head out of frame are still recommended as negative keywords.








About abysspastel:
you may want to repair the model according to this https://rentry.org/clipfix to get better outcomes.
Based off of the leaked model of you know what and trained on 60~ ish closeup of anime only, ????? only ???????. In short, the lora gives only ??????? which look like as in the preview pictures. If you want it in another angle/want a flaccid ?????, you should probably not use this.
Does basically what it should do, but sometimes has the tendency to move the head out of the picture. I recommend putting "head out of frame" in the negatives. Also, all four trigger words "?????, frenulum, ?????????, ?????" are recommended. "????????" may also be useful.
New dataset, suggested weight is ~0.4 or heads may disappear. close-up and head out of frame are still recommended as negative keywords.
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
