此模型的原作者为「@gennadiy」，LiblibAI 致力于推动原创模型的分享和交流，因此转载了该模型、供大家非商用地交流、学习使用。请使用该模型的用户，遵守原作者所声明的使用许可。

如果您是该模型的原作者，请与我们联系（邮箱：liblibai@163.com；微信：Liblibaijiang），我们期待您入驻、并将第一时间把模型转移至您的账号中。如您不希望在 LiblibAI 分享该模型，我们也将遵循您的意愿在第一时间下架您的模型。

LiblibAI 尊重每一位模型原创作者，更期待与每一位模型作者共同成长！

声明：若该转载模型引发知识产权纠纷或其他侵权行为，LiblibAI将立即下架模型，并不会向原作者追责。

Recommended weight 0.65-0.95. Interesting results with more weight, but more artifacts.

Sometimes there are problems with generating hair flaps and white sandals, solved by increasing the weights and reroll.

My first lora, so it's not ideal at all. It was noticed that sometimes one image from a dataset can be used too often. I haven't tried it with other loras, but I don't think the result will be bad.

Triggers: saya, saya no uta, hair flaps.
Maybe in the future I'll add images from the visual novel for the official artstyle.
Trained on Anything v4.5

---------------------------------------------------------------------------------------------------
V2!
In this version I adjusted captions via wd captions, which is why I strongly recommend to use v2 and not v1. I added support for visual novel style via official artstyle, official art, game cg tags. Also try nsfw (included in dataset), eldritch abomination, wings.

V2!
In this version I adjusted captions via wd captions, which is why I strongly recommend to use v2 and not v1. I added support for visual novel style via official artstyle, official art, game cg tags. Also try nsfw (included in dataset), eldritch abomination, wings.