Lora that adds sheer panties to Qwen. Should work in text to image as well as Qwen image editing.

v1.0

Based on 43 ai generated images

Trigger words: sheer panties

real V1.0

Based on ~250 real images

Trigger words: bodysuit, sheer panties

Though just "panties" seem to be affected as well.

I got best results on LoRA strength 0.9 and cfg 1.5 but YEMW

On both version I followed suggestions to remove background and faces on images before learning. Though I've seen other opinions, that suggests not to do that. Will try to learn simila LoRA and background/faces left intact. Also not clear how learning AI handles alpha channel on .PNG image and how to specify transparent color to caption. I am learning.

These are my intial loras so any feedback and suggestions would be appreciated