Ver.2.0 Train with more close up photo, the face look more accurate.
Mao_Hamasaki.safetensors
595063_training_data.zip