好累,我要用半年工资买4090……
建议权重在0.6~0.8使用,推荐0.6~0.7
The right of suggestion is 0.6~0.8, and 0.6~0.7 is recommended
不推荐:输入关键字“adawang,choker,”或 "fengsheng,military, military_uniform," 可能会触发某些场景或服装,但效果一般,我会继续更新
not recommend:Entering the keyword "adawang,choker," or "fengsheng,military, military_uniform," may trigger certain scenes or costumes,But the overall effect is so-so, so I'll keep updating
经过培训,仅用于测试。素材训练中参考了多人特征,非现实中真实人物,请勿对号入座,素材来自网络
Trained for testing only. The original material refers to a number of figures, please do not fit in, source network
稳定扩展模型:chilloutmix_NiPrunedFp32Fix.safetensors
所有训练图片均来自网络。
请不要在广告中使用此模型!
仅限学习交流用,用这种模型生成的图片自行承担风险
All training pictures are from the Internet.
Please do not use this model in advertising!
For learning and communication only, use this model to generate images at your own risk
学习コミュニケーシェン用ニ限しテ、このモドルド作成した絵でリスクを取ります。
Description
FAQ
Comments (10)
非常好!请问多少张图炼的?18+又是怎么炼的?
每个Lora基本上100张左右,要收集素材、剪裁,后期修图(尽量提升画质,比如调整亮度,白平衡,降噪等),还有很费力的是打TAG,18+我还是觉得有些玄学,个人建议是选择穿比较少的照片会比较有效果(当然原图就18+的更好)
为什么我不能复现LBB的容貌,加上关键词也不行
一般来说,多试几次,迭代步数调高,Lora在0.7左右,我觉得就可以(不像的话,不要用fengsheng,adawang这些关键词)
4090砍了,建议等5090
大佬来个赵今麦
来个大嫂 陈书婷
大佬求个范冰冰
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.