LoRA model trained on Flux.1 Dev. Trigger word: natann. Use strict settings: weight 1.0, steps 20-30.