Brief Introduction：
Only need 'yangyang', '1boy' as trigger words when using this version.
Don't use 'solo', 'furry', 'wild boar' in most cases, they cause random effect.
Recommend weight: 0.8-0.6
Known issues:
In certain heavily tuned non-furry base models, using only the tag '1boy' tends to generate a young human male, causing this LoRA character to lose its unique facial features. To fix this, you can try adding 'pig nose, tusks, mature' to the positive prompts, and 'human, young' to the negative prompts.
Full Introduction：
It's recommend to use '1boy, yangyang' as character tag since they were applied in model training, better not use 'solo', 'furry', 'wild boar', etc. Any character tags except '1boy' and 'yangyang' cause random effect, but of course you can try them.
Adopting distinct labeling strategies, this version only requires the single trigger word yangyang, delivering superior generalization, aesthetic quality, and facial features, as well as better compatibility with base models and other art styles. However, using tags other than yangyang and 1boy (e.g., furry, wild boar) will alter the character’s features (random hairstyles, facial details).
This version supports two usage modes:
1. Stable Mode: Set weight to 0.8 (above this causes overbaking). Use only the trigger words '1boy, yangyang' as character tag for the most faithful recreation of the original character design.
2. Generalization Mode: Set weight to 0.6–0.7. Only 'yangyang' is required as the trigger word, with other prompts freely customizable. Tags or phrases such as 'anthro wild boar with long black hair' can be used to stabilize character features.(Recommended for scenes containing non‑male anthro characters or blended non‑anime art styles.)
PS:
1. Whether the randomness at low weights is desirable depends on whether you prefer the original character design or only the core concept of the character. Lower weights create more variety.
2. If used alongside other LoRAs for optimal results, you may need to adjust their respective weights.
