LORA trained on 50 publicly available images of Saori Hara.

Trained on base SD 1.5, should work with most realistic models.

Use ADetailer if details are lost and strength of 1 should work on most prompts.
Will produce nudity without negative prompts.