Artist style: Uaxa

This Lora is a little bit overfitted because of very limited training data (only 7 images), you need to specify the features of the face to avoid always generate the same face. Nevertheless, I believe it is still worth a try.

由于非常有限的训练数据（仅7张图片），这个Lora有一点过拟合的迹象，在使用时需要手动指定脸部特征以避免总是生成同一张脸。