Use at 1.0 weight. See preview images for example prompts.

A few other people have made Kiss-Shot LoRAs. I already had this dataset from my 9-in-1 Kiss-Shot/Shinobu Oshino LoRA and I wanted to make a standalone Kiss-Shot LoRA that:

1. Doesn't have the character bleed of a multi-concept LoRA

2. Is trained with settings I know guarantee a certain degree of quality

3. Is guaranteed to work well with NAI-based model mixes and style LoRAs

4. Isn't 128dim

I hope you enjoy!