• Base Model: Anima Official

  Training Steps: 3000 steps (Rank 32 / Alpha 32)

  Mandatory Prompts: (kei, white hair), kei (blue archive), AnimaKei
  (Note: Please use "white hair" together with the trigger words to ensure consistent results.)

  Recommended Weight:

  • 1.0 (Standard)

  • 0.6 - 0.9 (Recommended when changing outfits; not recommended to go below 0.6)

  Training Method:
  This LoRA was trained using a mix of Natural Language captions to enhance aesthetic quality and overall visual appeal.

中文

• 基礎模型 (Base Model): Anima Official

  訓練參數: 10,000 steps (Rank 32 / Alpha 32)

  必要提示詞 (Mandatory Prompts):
  (kei, white hair), kei (blue archive), AnimaKei
  (註：請務必搭配 "white hair" 以確保角色特徵的一致性)

  建議權重 (Recommended Weight):

  • 1.0 (標準推薦)

  • 0.6 - 0.9 (換裝時建議調低，但不建議低於 0.6)

  訓練方法:
  本 LoRA 採用自然語言 (Natural Language) 標記混合訓練，能提供更佳的審美質感與視覺表現。