中文:
之前发布的 LoRA 因内容问题被平台下架,目前已重新上传并上架。原版本仍在申请恢复中!此前发布的三个版本均已整合在当前发布页面中。
English:
The previously released LoRA was taken down due to content-related issues. It has now been reuploaded and is available again. The original release is still under appeal for reinstatement. All three previously released versions are now available under this listing.
# Qwen-Image-2.1 NSFW Image Edit LoRA
全裸编辑 / Nude Edit · 开腿 / Legs Spread · 女上位 / Cowgirl · 口交 / Oral
给 Qwen-Image-2.1 图像编辑 用的 LoRA。
# Qwen-Image-2.1 NSFW Image Edit LoRA
Nude Edit / 全裸编辑 · Legs Spread / 开腿 · Cowgirl / 女上位 · Oral / 口交
A LoRA for Qwen-Image-2.1 image editing.
Give it a source image plus one Chinese instruction: turn a clothed photo into a nude, or into a legs-spread / cowgirl / oral edit, while keeping the face, hairstyle, accessories, and original composition as much as possible.
NSFW. Trained for undress to nude, not for normal outfit swap.
Description
FAQ
Comments (24)
要不你把人物头像打个码。展图换成3D角色好一点,毕竟这个东西,一点涉及真人,很容易被禁掉
Maybe blur the avatar/headshot first. It would be better to change the display image to a 3D character, because once real people are involved, it can get banned easily.
原图也是k2生的,这次还被下架的话我试试别的,排名第一的那个lora也是真人图
@oreolove 如果太像你可以注明是用什么生成的,尤其是太像真人的。
If it looks too realistic, you can note what method was used to generate it, especially if it resembles a real person.
@wwwmanmoon103 再看吧
为啥下架啊?你发童了?
涉及到编辑脱衣了,童的话只会把涉及的图片删了
@oreolove no, not because it involves editing nudity. you chose very questionable images in the previous version you uploaded. I don't think the previous upload will get reinstated. it's a great lora by the way.
@napsumfur225 Ah, I see. So that's how it is. Thanks for explaining!
和新出的turbo 8步模型貌似相性不是很好,和原版40步比效果差很多诶
虽然turbo配合不是很好 但是质量很顶! 配合ray当enhancer其实能超过krea2 enhancer(当然也可能是我的姿势不对) 当前lora配合raw,enhancer配合turbo可能就是最终答案 期待老哥迭代
可能需要重新练,turbo需要独立炼lora
@oreolove 对 这个是个问题 然后还在测试一下2K场景下的性能 目前看1K是很稳定的 eular_a 配合 liner 锁脸也不错,当然还是得配合Ray洗一下。 现在已经很厉害啦,期待迭代
@Dr_BB_G V2的一致性可能高一些,后面出的应该没那么高了,加了很多变化大的数据会对一致性有些影响
@Dr_BB_G 2k的话我turbo练好之后试一下
@oreolove 我的解决方案是 2nd pass 更换 diffusion (Ray那个版本) 然后用linear schedular,锁脸是可以的。目前只能1st pass争取把所有概念做到位,2nd pass 做修复和脸部匹配。 turbo做修复优化可以,但是编辑感觉能力还是差太多了
@Dr_BB_G turbo只能cfg1是差点,原版都要开到3才好用
@oreolove 不知道老哥 有没有测试过ray版本的diffusion,我还在找和这个lora匹配的参数,毕竟base变了,权重还没有找到合适的点
@Dr_BB_G 没,我肯定用官模试的
大佬快出turbo版本的,turbo版本感觉配合并不是很好。
V3 很完美;角色的逼真度和一致性得到了提升,皮肤质感也很棒。它还修复了生殖器畸形的问题,并且补全了此前缺失的肛门。肢体缺失的问题也得到了显著改善。感谢作者。



