已上传训练集，需要PSD，jpg及txt原始文件请联系我，B站：鱼泠溟；私信我，暂时只向官方，其他lora训练者或认为我使用了您作品的画师提供
The training Data has been uploaded
提示词：Katya,1girl,
额外可选提示词：dress,flat chest,loli,
跑图可以使用DPM ++2M或DPM ++SDE算法
权重根据模型不同可以在0.6-0.8左右调整，权重越大鸭鸭越像，但生成纯白背景的概率越高,
可以配合hire fix使用，放大算法请选择为R-ESRGAN 4x+Anime6B
第一次训练过程使用了40张在星律动官方公布的3d模型基础之上，由自己修改后模型生成的图片，第二次训练过程使用了40张由AI生成和自己瞎画的图像，版权归NB厂（星律动）所有,
严禁商用，严禁公开传播由此lora生成nsfw图像及做出任何损害鸭鸭名誉权的行为，如有冒犯，请联系我删除此Lora.
Associated prompt words: Katya, 1girl,
Additional optional prompt words: dress, flat chest, loli
The DPM++2M or DPM++SDE algorithm is recommended for running the map. The performance is pretty good with a weight of 0.6-0.8;
It is recommended to use it with hire fix. Please select R-ESRGAN 4x+Anime6B for magnification algorithm.
The first training process used 40 3D models published by Nebula Beat, and the second training process used 40 images generated by AI. The copyright belongs to Nebula Beat.
Commercial use is strictly prohibited. It is strictly prohibited to publicly disseminate nsfw images generated by this lora and make any behavior that damages Katya’s reputation.If there is anything wrong, please contact me to delete this Lora
描述:
这个版本的鸭鸭（可能）会更像
但
极大概率（99.99999%）背景会强制锁定为纯白（更好抠图了），有特殊需求可以下载这个版本
训练词语: katya,1girl
名称: KatyaV21.safetensors
大小 (KB): 147571
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
