Sorry to see the SDXL version was deleted but I saved a copy locally and am hoping to try them out this weekend, once I figure out runpod/vast.ai/thinkdiffusion/rundiffusion. I commented on that one how the body looked spot on but the face was "merely" OK and wondered what the ratio of body:face pics in the training data was but it was deleted before I could read your reply. I guess I'm just curious about the datasets/tagging between the two and wondered about any of your thoughts post-training.

Anyway, I'm looking forward to trying them out. Thanks!