封面为混合风格(我知道这样不太好但实在好看!)

The cover features a mixed style (I know it's not good but it looks really nice!)

以doll和空镜头为素材进行训练,在768分辨率进行试验,旨在为风格融合提供真实质感但脸又不那么大妈!

Conduct experiments at 768 resolution and apply dddolls and empty shots as training materials to provide semi-realistic texture for style fusion while keep the facial features of big cute eyes

内置触发词@dddoll,视乎情况决定是否使用

推荐与下面的lora进行混合使用,他们有相近的标签处理意图,推荐强度在0.3~0.6之间

It is recommended to use them in combination with the following LoRAs, as they share similar intentions for tag processing.The recommendation intensity ranges from 0.3 to 0.6

https://civarchive.com/models/2563437/semi-nffa-style

https://civarchive.com/models/2605893/3d-skin

已知问题

1.对初音未来和oni girl有过度响应Excessive response to Hatsune Miku and Oni Girl

2.手指有时候特别短Fingers appear unusually short occasionally

3.球关节的表现暧昧,尤其是胯部The performance of ball joints is ambiguous, especially at the hips

4.dddoll的眼型没有作进一步区分The eye shapes of dddoll are not further classified.