Note that this LoRa is very experimental so it requires many trials and errors because it used only four images for training. This might require to use with fumio LoRA.