My MOSS Character Lora Training Results

Hi all! What do we think of MOSS Image so far? Personally I'm loving it. The variety in images? The differences in people's faces, ethnicities, body types, etc? And it's so easy to get my favourite style of photo to create, amateur candid realism. It really is a great model. Lora training was kind of a tough nut to crack though, but I'm pretty certain I have settings that work, so I figured I'd share. With these settings, I get good results every time. Most of my datasets were between 50 and 100 images, with natural language captions. Best settings I found are Automagic3, with Sigmoid and High