Trained by my painting, have fun! (ﾉ∀`*)
I think it is some kind of Gacha or lottery, DRAW!
俺のターン、ドロー！
Some base models may need a LoRA strength of around 0.7–1.6 for best results.
If you use Tsubaki2, you might use 0.5-0.7, or you will get some clumsy hand-painted. (my fault, still renew this LoRA)
This LoRA reflects my personal style, so the outputs may lean toward:
- Sidelocks, messy or curly or layered hair, ahoge
- Noticeable reflections or a soft airbrushed look
- Without eye highlights or glowing lower eyes
- Stocky or shorter proportions, occasionally cartoon-like
It tends to work best for tachie-style art or avatar illustrations.
用自己以前畫的圖訓練的LoRA，畫風有點雜，在不同模型會有截然不同的成果，大概是從資料大混池裡面抽一張圖出來參考。
適合的LoRA強度大概在0.7~1.6之間，依模型而定，目前在邊玩邊看會冒出什麼，有喜歡的畫風可以直接套用。
如果是用Tsubaki2，強烈建議權重使用0.5~0.7，否則會拿到很多歪歪扭扭的圖，還在改進中。
因為本人的繪圖習慣，你可能會獲得：
- 鬢角、亂髮、捲髮、多層次頭髮、呆毛
- 隨性的噴槍打光
- 眼睛無高光但眼睛下部帶著亮光
- 角色畫起來好像會偏矮胖，或是有點卡通化
大概是一個適合用來繪製立繪或頭像的LoRA，持續測試中！
