"姐姐我啊，最讨厌花言巧语的男孩子了~"脸部是根据现有的lora调权重得出的，同时喂了一些淘宝的衣服模特图和p过脸部的vam图，可以生成出比较多样的衣服和动作，泛化能力还不错。只用头的话加yingHead，用衣服和姿势加yingBody“As a sister, I really dislike the boys who always use honey words~"The face was generated by combining multiple individuals' loRA models, and incorporating clothing models from Taobao and posture models from Virt-a-Mate, resulting in a diverse array of clothing and movements with good generalizability. To generate just a head, use yingHead, and for clothing and posture, use yingBody.

Files with SHA256: c3d5ba3a2f0eec3ac48a7ac2f62032b0007513effa7a3439ad4a83efd574bab4