Reshiram LoRA trained on NovelAI with Kohya training scripts using the DAdaptAdam optimizer at 512x512 resolution. This one is very spicy, I recommend you run it at a lower strength (0.5-0.8) if your base model already has an understanding of Reshiram's appearance. I added (metal bands) to my negative to help with the excessive amount of metal bands produced on higher LoRA strengths. Tail placement is inconsistent. Despite the issues, good results are possible. Should have wide compatibility with model mixes based on NovelAI, which is most of them. Trained on a mixed male/female dataset.