Quick Generate
Select a style, write what you want to see and click Generate
best quality,masterpiece,highres,realistic,1boy,upper body, looking at viewer,city, rucksack, white shirt, black-framed eyewear, <lyco:niji (2):1.0>,niji3,
best quality,masterpiece,highres,realistic,1boy,upper body, looking at viewer,city, <lyco:niji (2):1.0>,niji3,pectorals, abs, biceps, white hair,fake facial hair,bandaged arm,sweat,
What you want in the image.
Clear
Random

Model Description by Creator

禁止使用此模型进行一切商务及违法行为,禁止随意转载,仅作为成果分享,违者后果自负!

我在国内站tusi.artliblibai.com里也都有模型,按平台的创作激励会有所不同,还请多多关注支持!

(我的QQ群:235392155,Lora代练、ckpt融合调整加我qq:2402799912)

【闲鱼】https://m.tb.cn/h.5V7ITvv?tk=eY01dB4UcnH

这是我的爱发电,补贴一下电费QAQ,感谢您的支持~

https://afdian.net/a/_Qing_

V2:

同训练集,调整训练参数,推荐权重为1

V1:

利用niji跑图练的风格lora,除本身画风外。还能提高对男性塑造。推荐权重受大模型影响,可自行尝试

给予访客使用模型的一些建议与回答

(作为一个外行人,虽然没有经历过美术和ai的专业课学习,但我仍然在不断努力学习探索,并与各位大佬交流心得。我对自己在意的事情有着很高的精神洁癖,因此我希望能在自己的一亩三分地里传达正确的ai知识。)

1、LORA

  • 对于动漫角色lora而言,最好的权重便是1。虽然我们能通过改变权重的方式来改善拟合,但改变权重会额外产生一些不好影响。比如:c站动漫角色lora大多数都过拟合,而为了改善这一现象往往会推荐降低权重到0.6/0.8,而这一举动会导致角色失去部分原本的外貌特征

  • 过拟合通俗解释:过分训练导致的lora表现僵化,如人物不听tag话甚至出原本素材的图

2、CKPT

  • clip偏移和unet过拟合都会导致模型不听tag话

  • clip偏移会导致tag识别问题,c站有相当一部分模型存在此问题而不自知,如果你感兴趣,在下面的“其他”里我会给予检查及简单修复的途径

  • unet过拟合也会让模型僵化,不听tag使唤,甚至在无tag的情况下跑出好看的图片,这也是过拟合吐原图的情况

  • 融合类ckpt能够很好的提高模型出图的下限,但融合模型对于tag的权重是很乱的,因此会各具“特色”,同样,基于此类模型训练的lora很难在其他模型上适用

3、VAE

  • CKPT自身就有一个VAE,外挂VAE不是用于额外的增加,而是用于替换

  • VAE只是最直观的是饱和度变化,这并非VAE功能的全部,就对于跑图而言还会改变构图、细节等等

4、其他

  • 模型占内存大的不一定好,很多模型内有一堆无用数据,下载完全浪费流量内存

  • 示例图的好坏不能确定模型质量,除了基于作者本身审美外,你无法确定他用了多少模型、插件的辅助,你也无法确定他文生图跑了几次、图生图跑了几次

  • 下载量与点赞也不能确定模型质量,点赞量基于下载量,而下载量会和作者名气、封面图片受众、模型角色人气、模型画风受众、模型类型受众等等相关

  • 模型并非版本最新为最好,很多时候是基于某一版本的不同方向的调整。甚至有部分作者为了通过更新的推送而蹭到更多的下载量而恶意刷更新,实际使用下来模型的质量不过是在原地迂回踏步

  • 其他更为专业详细的知识请看万象熔炉 | Anything V5/Ink的简介部分,关于clip偏移的检验修复的扩展链接在此文档前段部分中

    模型链接:万象熔炉 | Anything V5/Ink - V3.2++[ink] | Stable Diffusion Checkpoint | Civitai

Images Generated With This Model

best quality, masterpiece, highres, 1boy,male,solo,male focus, upper body, <lora:niji3:0.5>
N/A
Hidden For Public Users
best quality, masterpiece, highres, 1boy,male,solo,male focus, upper body, <lora:niji3:0.5>
best quality,masterpiece,highres,realistic,1boy,upper body, looking at viewer,city, rucksack, white shirt, black-framed eyewear, <lyco:niji (2):1.0>,niji3,
N/A
Hidden For Public Users
best quality,masterpiece,highres,realistic,1boy,upper body, looking at viewer,city, rucksack, white shirt, black-framed eyewear, <lyco:niji (2):1.0>,niji3,
best quality,masterpiece,highres,realistic,1boy,upper body, looking at viewer,city, <lyco:niji (2):1.0>,niji3,pectorals, abs, biceps, white hair,fake facial hair,bandaged arm,sweat,
N/A
Hidden For Public Users
best quality,masterpiece,highres,realistic,1boy,upper body, looking at viewer,city, <lyco:niji (2):1.0>,niji3,pectorals, abs, biceps, white hair,fake facial hair,bandaged arm,sweat,
1girl, CCCPposter, sovietposter,none,soviet poster, soviet
N/A
Hidden For Public Users
1girl, CCCPposter, sovietposter,none,soviet poster, soviet
1girl (watercolor:1.3),vivid color, <lora:den2SD:0:NONE>
N/A
Hidden For Public Users
1girl (watercolor:1.3),vivid color, <lora:den2SD:0:NONE>
1boy , outdoors,upper body,red hoodie,piec, <lora:den2SD:0:NONE>
N/A
Hidden For Public Users
1boy , outdoors,upper body,red hoodie,piec, <lora:den2SD:0:NONE>
1boy , outdoors,upper body,red hoodie,piec, <lora:den2niji:0:NONE>
N/A
Hidden For Public Users
1boy , outdoors,upper body,red hoodie,piec, <lora:den2niji:0:NONE>
1girl (watercolor:1.3),vivid color, <lora:den2niji:0:NONE>
N/A
Hidden For Public Users
1girl (watercolor:1.3),vivid color, <lora:den2niji:0:NONE>

Train Models

Train AI Models On Your Own Pictures

You can use your own pictures to train AI models. Reimagine yourself in new ways with our AI-powered image generator.
Create pictures of yourself around the world, in your dream outfit, as a video game character, or in countless art styles. The possibilities are endless.
Train Your Own Model

Generate AI Images. Simple and Easy

AIEasyPic is the #1 AI image generator. Generate images with simple prompts, swap faces, train models to reimagine yourself, and more.