Articuno LoRA trained on NovelAI with Kohya training scripts using the DAdaptAdam optimizer at 512x512 resolution. Trained on a dataset with predominantly 'normal' arms instead of wings or winged arms. You can get good results by adding (winged arms) to your prompt, but it can take a while to get an output that looks good. Often misses part of her head crest when it's off axis. Her tail 'ribbon' can sometimes turn into a feathered tail. Should have wide compatibility with model mixes based on NovelAI, which is most of them. Trained on an all female dataset.
