SD1.5模型下的lora,用yesmix模型,

偏向动漫和现实中里的中间的绘画风格。适合生成御姐风女性及NSFW生成，权重不高于0.8最好，可以融合角色lora一起使用。

lora under the SD1.5 model, use the yesmix model,

I prefer a painting style that is halfway between anime and reality. It is suitable for generating ladies with royal sister style and NSFW. It is best if the weight is no higher than 0.8. It can be used together with the character lora.

使用方法Instructions：

角色lora过高可能会出现画面异常，建议角色lora和本风格lora以负相关调节。

同时放大功能里重绘幅度不宜太高，0.4——0.7基本合理，过高重绘幅度会导致肢体异常。

建议开启ADtailer，修复的脸部能保证基本完美。

与controlnet配合度较高，可以多采用controlnet控制姿势

If the character lora is too high, the screen may be abnormal. It is recommended to adjust the character lora and the style lora in a negative correlation.

At the same time, the redraw amplitude in the zoom function should not be too high. 0.4-0.7 is basically reasonable. Too high redraw amplitude will cause limb abnormalities.

It is recommended to turn on ADtailer, and the repaired face can be guaranteed to be basically perfect.

The degree of coordination with controlnet is high, and controlnet control postures can be used more often

模型优点advantages：

风格独特（我个人比较喜欢），生成女性气质独特且身材好

擅长生成全身图，以长图为主要。

脸部和头发绘制细致，色彩鲜艳

适用NSFW生成

Unique style (I personally like it), generates unique feminine temperament and good figure

Good at generating full-body pictures, mainly long pictures.

Face and hair are drawn in detail and colorful

Suitable for NSFW generation

可能的问题possible problems：

手部异常，手指的生成效果一般。

动作偏少，对动态姿势兼容性一般

The hand is abnormal, and the finger generation effect is average.

There are few movements and the compatibility with dynamic postures is average