lora有水印，所以反向提示词里记得添加“水印”。

这个lora里有面部，所以与其他lora混着使用时，最好使用lora分层插件，

看看lora在不同权重下的表现

当步数在28时，推荐权重0.4～1

本人电脑性能有限，大家可自行测试。

lora训练基础参数

大模型：chilloutmix

vae:vae-ft-mse-840000-ema-pruned

采样方法：DDIM

There is a watermark in the Lora, so remember to add the word "watermark" in the reverse prompt when using it with other Loras. As this Lora contains a face, it is recommended to use a layered Lora plugin when mixing it with other Loras to see how Lora performs at different weights. When the step number is 28, a weight of 0.4 to 1 is recommended. Please note that my computer performance is limited, so you can test it yourself. The basic training parameters for Lora are as follows:

Large model: chilloutmix

VAE: vae-ft-mse-840000-ema-pruned

Sampling method: DDIM.