Just one-shot trained LoRA on bad small dataset. A lot of work to do in next ver.

Very overfitted, face problems, blades are random disaster. Trained on Anything3.0

Check prompts in pictures info.