User LoRA
https://civitai.red/models/2290713/stylekinoshitakou-style?modelVersionId=2662097
1.1版本/ver1.1:推荐强度1.0,根据自己出图效果调整优化了训练集,把带文字对话的图全部去掉再训练了,更稳定.推荐强度:1.0~1.3不需要正面提示词负面提示词推荐:speech_bubble, sound_effects, logo, text