v1. Trained on 196 images. Epoch #8 chosen for best resemblance. Options #10, #12 retained for consistency testing.
Note: LoRA works horribly for european checkpoints...
v1. Trained on 196 images. Epoch #8 chosen for best resemblance. Options #10, #12 retained for consistency testing.
Note: LoRA works horribly for european checkpoints...