a little bit more detailed due to the larger dataset v1=30-40 img, v2= 200 img
raw version of lora of the character a The Herta
