Trained on pupi artist dataset. Around 200 images, most were portrait so it is suggested to generate in that format. Most images had women, so lora tends to skew towards that but it can generate feminine men as well. Has a focus on pantyhose and feet.