=3=

第一次训练时，原本想制作一个能生成 =3= 表情的 LoRA，但人物本身也大量被训练进去了，基本没有实用价值（只能生成真人风格的图像）……

由于素材有限，基本只能生成 512×512 的上半身或特写等正面或侧面人像。

推荐权重 0.9，使用 closed mouth 生成嘴型，closed eyes 生成眼型。

First time making a Lora, was supposed to make a Lora that makes the figure with facial expression of =3=, but Yuuma takes a great proportion in it as well.

due to the limited material, basically it capable of producing only 512*512 closeup/upperbody portraits

rec: 0.9 weight, prompt: Closed mouth, Closed eyes,