训练集是ai的图片,有些手有问题，模型应该使用一个手部生成效果好的

The training set is a picture of the ai, some hands have problems, the model should use one hand to generate good results

它在512x768训练，使用更高的分辨率获得更清晰的效果

It trains at 512x768 and uses higher resolution for sharper results

这个lora过拟合非常严重，如果你没有经验，请参考我的lora权重以及每个分层的设置。

我花费了很多时间来调试，让画面变得真实一些。

This lora overfit is very severe, if you are inexperienced, please refer to my lora weights and the Settings for each layer.
I spent a lot of time tweaking it to make it realistic.

【(mypeephole:1.3)】调高这个权重使画面更扭曲 Increasing this weight makes the picture more distorted

【breasts, navel,sexy nipples, pubes,underwear, ,bangs,jewelry, black pulic hair,indoor,slim body,shiny skin,soft light, ray trace,depth of field,

crowd】对于这些参数你可以自定义，它不是必需的 You can customize these parameters, it is not required