This LoRA is very rough and experimental; It is also the first LoRA I’ve created using Prodigy optimizer. Expect very little without specific prompting or inpainting.
It was very difficult to create a dataset for this character, so the LoRA’s performance may be hit or miss. Trained on FluffyRock, but tested with varying results on other furry models. Works best on models trained on e621 tagging.
This LoRA was not trained with NSFW in mind, nor is it in the dataset, so if you try to do so, expect mixed results.
Activation keyword is nanigashi; LoRA weight should be around 0.8 ~ 1, but the keyword itself may need its weight adjusted in some cases.
Recommended positives: 3 fingers, big hands, yellow eyes, white coat, 2 horns, semi-anthro (Add mask, feral, and belt if necessary)
Recommended negatives: anthro, feathered hands (Sometimes beak tags might be necessary, especially if the mask is being stopped from covering the entire front of the face by a generated beak)