
不知火舞这个lora身材高度还原,并且替换了个人更喜爱的脸型,权重设置在0.6~1之间,lora模型以墨幽人造人1010为基底炼制,训练底图大小768*768,,触发词为shiranui_mai
The lora figure of Shiranui Mai is highly reproduced and replaced with a face shape that is personally more favorite. The weights are set between 0.6~1, the lora model is refined with 墨幽v1010 as the base, the size of the training base image is 768*768, the trigger word is shiranui_mai
描述:
Positive prompt:absurdres, highres, ultra detailed, (1 young beautiful girl, solo:1.1),shiranui_mai,ponytail, red clothes,brown_hair , brown_eyes, high_ponytail, breasts, large_breasts, japanese_clothes,sleeveless,rope, long_hair,
Negative prompt:EasyNegative, (worst quality, low quality:1.4), [:(badhandv4:1.5):27],
参数选择Steps: 30, Sampler: DPM++ SDE Karras, CFG scale: 7,高清修复采用 R-ESRGAN 4X+ Anime6B,LORA权重建议0.6~1,权重为0.8即可。
Parameter selection Steps: 30, Sampler: DPM++ SDE Karras, CFG scale: 7, R-ESRGAN 4X+ Anime6B for HD repair,LORA weights are recommended to be 0.6 to 1, and a weight of 0.8 is sufficient.
训练词语: shiranui_mai
名称: Mai_uwuxo.safetensors
大小 (KB): 147567
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success