Anime lora trained on NAI. I do not think it is compatible with realistic models……

基于NAI训练的二次元lora。我不认为它跟真人模型兼容……

This is a 2in1 lora, there are two main trigger words:

sandrone: basically only Sandrone.

sandronerobot: Sandrone and her big robot. It is not fully tested. Check the preview images.

这是个二合一lora，有两个主要的触发词：

sandrone：基本上只有桑多涅。

sandronerobot: 桑多涅和大机器人。并未经过完整测试。可以参考预览图。

There are too many Sitting Sandrone images in the dataset, and she is wearing so many clothes. So it is suggested tweak the weight of lora to 0.7~0.9, and add more weight to pose tags.

训练集里的桑多涅大多是坐着的，而且她穿得也太厚了。因此建议将lora的权重设为0.7~0.9，并为动作tag添加更多权重。

Suggeted Prompts: black_fur, white_gloves