Due to the large size of the LoRA, I'm remaking it 2025/1/19
We trained the model using 200 images with multiple views and tags of upskirt and from below.
LoRA weight: 0.7~0.9
You can create similar illustrations without using LoRA, but using only the trigger word tag will help you create a stable sample composition.
⬇other sample prompt⬇
black pantyhose
Skindentation
Feet out of frame
Squatting