V2.0基于Animagine XL V3.1训练的版本，仅适用于Animagine XL V3.1Training base on Animagine XL V3.1. Only applicable to Animagine XL V3.1.对训练集进行了进一步精选Further selection has been made to the training set.有着更好的人体，PVC风格更加稳定Has a more stable style and better body structure.V1.0PVC质感的二次元XL lora模型。An anime XL lora model of PVC texture.基于青叶老师的Kohaku-XL gamma训练，在Animagine XL V3上有比较好的效果，但在不同CKPT上使用效果可能有较大差异。Training base on Kohaku-XL gamma. There is good performance on the Animagine XL V3, but there may be significant differences in performance when used on different CKPTs.使用超过7000张PVC图片及700张bjd娃娃图片直接训练生成。Training using over 7000 PVC images and 700 bjd doll images.感谢dorioku老师与sumomo老师在素材收集上提供的帮助。

Files with SHA256: 463cb1ee8fe26b676ddd72279448da5458d5daf7d14684b5032548fb519c1ef4