v1.0头饰的数量形状会有一般概率不正确,拉链的特征学习的不是很理想。推荐权重:0.5~0.9推荐加入cleavage来达到更好的效果,笑The quantity and shape of the ornament will have a general probability of being incorrect, and the zipper feature learning is not very ideal.Suggested weight: 0.5~0.9Suggested adding cleavage to achieve better results :)v2.0尝试修复了下头饰得形状,但是感觉肢体上得错误变多了,搭配controlnet使用吧推荐权重:0.75Tried to repair the shape of the lower headdress, but I feel that there are more errors in the body. Use it with controlnetSuggested weight::0.75

Files with SHA256: 999b9ec821ff2d36816645f8acba79c017176da040a8161d35406ed3e8d00708