This is my first upload. Tried this a few times with no success. After pruning though the sample images I got something fairly good. I think I used a protogen model to train this. All the SD 1.5 models turned out awful. LORA was trained at 768*768.