Type | |
Stats | 178 2,979 |
Reviews | (36) |
Published | Sep 21, 2024 |
Base Model | |
Training | Steps: 696 Epochs: 10 |
Usage Tips | Clip Skip: 2 |
Trigger Words | yuumi-default blue fur, chest tuft, yellow whiskers, blonde eyebrows, headgear, necklace, yellow tail tip feral, animal, quadruped riding book, floating object anthro, biped |
Hash | AutoV2 FE3B14368F |
Preview images made with Nova Furry Pony V3: https://civitai.com/models/503815?modelVersionId=777603
My second LoRA training. This proved to be a difficult character to train. The trigger word is usually NOT enough to represent the character, and you'll have to add some reinforcement tags to get her likeness.
This model also has a tendency of generating more than two whiskers on Yuumi, I don't know what I would need to do for it to get the number correct.
Finally, due to the dataset, this model is biased towards NSFW, however it can still generate SFW images with the correct tags and negatives.
Trigger word: yuumi-default
Reinforcement tags: blue fur, chest tuft, yellow whiskers, blonde eyebrows, headgear, necklace, yellow tail tip
Feral version: feral, animal, quadruped
Riding on her book: riding book, floating object
Anthro version: anthro, biped
I also trained another version that only requires the trigger word to replicate her likeness, but is much less detailed and mixes several of her traits together into a mush. I'll test that one a bit more and depending on the results I might upload it later.