表紙が削除されました： コンテンツ規則違反
I also starting remade old tags and clean datasets for training universal one character lora model... Idk, when i made it completely.Best weight for me: 0.8Trained on AnimaYume28/07 Opened how increase image quility with training lora of multiple charcters, I had bad results with Celestine and Kaguya, but now it's good.Until making All-one Lora I will train and update in near feature other old models.
