Download
1 variant available
While the Anima base model can generate a beautiful Kafuu Chino, a Chino without Tippy on her head feels like she’s missing her soul. To fill that void, I created this high-purity Tippy LoRA to bring the guardian of Rabbit House back to life!
Base Model: Anima [Official] (SDXL)
Training Parameters: 2000 steps (Rank 32 / Alpha 16)
Trigger Word: tippy
Recommended Setup: Best used with character tags like kafuu chino or hoto cocoa. Use animal on head or holding animal to guide the positioning.
Weight Guide
This LoRA was trained with high purity (Final Loss: 0.0159). Please adjust the weight according to your needs:
0.1 - 0.3 (Mini Decor Mode):
Tippy appears in a smaller size, similar to a delicate hair ornament. This range has zero impact on the character's facial lighting, perfect for high-fidelity portrait shots.0.4 - 0.7 (Golden Range):
Recommended Value: 0.6.
Tippy appears in standard size with the most rounded, "mochi-like" shape. Perfect for character interactions (on head or being held) while maintaining natural facial lighting.0.8 - 1.1 (High Intensity Mode):
Extremely sharp features. However, higher weights may start to override the base model's facial lighting and details. Useful for specific outfits or stylized filters.1.2 and above (Pure Tippy Mode):
The character will disappear, leaving only Tippy in the frame. Best for generating standalone Tippy assets.
Testing Example
Prompt:1girl, solo, looking at viewer, kafuu chino, (tippy:0.6), animal on head, indoors, cafe background, masterpiece, best quality
Technical Note:
This LoRA was trained exclusively on "Pure Rabbit" datasets to prevent character leakage/contamination. If the character disappears at weights above 1.2, it is a normal behavior (indicating the LoRA successfully isolated the background and subjects). For character interactions, please keep the weight below 0.8.
以上google機翻英文
下方中文
雖然 Anima 模型已經能生成非常精緻的智乃 (Kafuu Chino)，但頭上少了提比 (Tippy) 總覺得少了靈魂。為了填補這份遺憾，我製作了這款高品質的 Tippy LoRA，讓 Rabbit House 的守護神重新顯靈！
基礎模型 (Base Model): Anima [Official] (SDXL)
訓練參數: 2000 steps (Rank 32 / Alpha 16)
觸發詞 (Mandatory Prompt):
tippy建議環境: 搭配
kafuu chino或hoto cocoa等角色標籤使用，建議加入animal on head或holding animal引導位置。
權重建議與效果指南 (Weight Guide)
本模型訓練純度極高（Loss 0.0159），建議依據需求調整權重：
0.1 - 0.3 (迷你裝飾模式)
Tippy 會呈現較小尺寸，像精緻的頭飾。
完全不影響角色臉部光影，適合極致追求畫質的場景。
0.4 - 0.7 (黃金推薦區間)
標準推薦值：0.6
Tippy 呈現標準尺寸且形狀最圓潤。
能與角色完美互動（如頂在頭上或抱在懷裡），且角色臉部光影保持正常。
0.8 - 1.1 (強力模式)
Tippy 特徵極其鮮明，但因權重較高，可能會開始搶走角色臉部的光影與細節。
建議換裝或使用特殊濾鏡時可在此範圍微調。
1.2 以上 (純粹提比模式)
角色會完全消失，畫面轉為純粹的 Tippy。
適合需要單獨生成提比素材時使用。
測試參考 (Testing Example)
提示詞範例：1girl, solo, looking at viewer, kafuu chino, (tippy:0.6), animal on head, indoors, cafe background, masterpiece, best quality
技術提示：
本模型採用「純兔子素材」訓練，不含人物污染。若在 1.2 權重以上導致角色消失屬於正常現象（代表 LoRA 成功過濾了背景與人物）。若要與角色互動，請務必將權重維持在 0.8 以下。