I did this LoRa experimentally, just for fun. Might have further adjustment.
隨便做做,有時間再改進這個LoRa
Base model: majicMIX realistic
Training based on 512x640 image
Recommended weight:0.7~0.9
meimei.safetensors