One obsession_Anima（Open free download）
https://civarchive.com/models/2695493/one-obsessionanima
Mature.性感 木珠 NoobAi.Vpred:
https://civarchive.com/models/1368727/one-obsession-branchmaturenoobaivpred
Please support my new NoobAI model: https://civarchive.com/models/2558457/oneobsession-hybridtheory-noobaiillustrious
请支持一下我的NoobAI新模型：https://civarchive.com/models/2558457/oneobsession-hybridtheory-noobaiillustrious
If you like my work, please give me a good review and click the thumbs-up icon. Please give me more feedback! It's free, and it can keep me motivated to create.
如果你喜欢我的作品，请给我一个好评并点击大拇指图标。多多返图！这是免费的，而且能让我保持创作动力。
You can make a good picture without lora, without complicated tags, and it is extremely friendly to beginners!!!!!
不用lora也能出好图，不用复杂的tag，对新手极其友好呦！！！！！
Still, I repeat, the new version may not necessarily be the best. Experience it yourself and have fun!
^^!绘画标签都藏在图片提示词中！寻找画师串也是乐趣所在！例图未加任何lora，其他自己探索吧！还是那句话，各有特色新版本不一定是最好的。自己体验，玩得开心！
AI绘画QQ交流群：
（在用）点击链接加入群聊【痴迷AIGC备用群】：https://qm.qq.com/q/XZutAVZVei
点击链接加入群聊【痴迷AIGC生活继续新新新1群（第三次转生）】：https://qm.qq.com/q/Uqw8uS7I0C
点击链接加入群聊【痴迷二次元AIGC永远不说再见之新新新新1群（第四回）】：https://qm.qq.com/q/EdWjbCj1Bu
点击链接加入群聊 少女宗所在群，这里不欢迎木珠 【痴迷AI新3群二次元少女宗】：https://qm.qq.com/q/DkcPLAh1uw
请认真回答从哪来？瞎回答的不予通过！
感谢QQ扩群的群友募捐： 木珠门大长老 夏雨声烦 antanc 109°28’ duiker 普通男性 摸鱼仙人 夜小钗 风纪委员 pipiai 天子doro paff 飘忽 夏沫 苏沐凝萱 牢冰 China头子 underlzy 锐子 库里垮炎 pixard 沉默的威斯东 多出来的几十块推到下一年，还缺谁Q我！
前三个个群如果都搜索不到就进DC，毕竟DC&TG可以涩涩
Discord：https://discord.gg/7RNWedD59H
tg：https://t.me/+ar2q0dOvGVAxZDA1
^^! The painting tags are all hidden in the picture hints! Finding artists to collaborate with is also a lot of fun! The example picture hasn't been enhanced with any lora, and the rest is up to you to explore! As I've said before, the new version may not be the best. Experience it yourself and have fun!
Discord：https://discord.gg/7RNWedD59H
TG:https://t.me/+ar2q0dOvGVAxZDA1
推荐一个forge或webui的插件，方便 Illustrious的动漫角色的选择，插件可以扩展URL安装，插件链接如下：https://github.com/lanner0403/WAI-NSFW-illustrious-character-select
可盐可甜可干爽可油腻可3D可2D 可涩涩可清纯 打造泛用性最强二次元Illustrious微调大模型！（狗头）
光影方面建议加：(chiaroscuro,high contrast:0.5),(sunbeam), 等来强化光影
如果希望画面变暗可加： (dark,chiaroscuro,hight-contrast:1.7),
人物半写实可加：((realistic, photorealistic, realistic shading)),
还原动漫可以加：(anime screenshot),(anime coloring ), 动漫截图，动漫上色，来还原
2.5D半写实画风：(photorealistic, realistic, 3d, photography),
2D画风：((flat color:1.8)),(abstract:0.5),只用前面的平涂不带后面的抽象也可以。也可以再加上：(anime screenshot),动漫截图，(anime coloring ),动漫上色。
关于画风的绘画标签，画师串太多，自己研究吧！
The thief website [PixAI] has illegally used my model without my permission. Moreover, my model is free and has been published on...
小偷网站[PixAI]，未得到我的许可，非法使用我的模型。而且我的模型是免费的，发布在
★NOW★
推荐分辨率：1024X1536 、832 x 1216、896 x 1152、768 x 1344、640x 1536、
recommended resolution：1024X1536、832 x 1216、896 x 1152、768 x 1344、640x 1536、
V21动漫实验版本：质感更润了，加强了光影，弱化了一些3D，人物偏可爱，对画风，画师串敏感；画风兼容性不错，有些画风会有点抽象。
Anime experimental version, texture is more moist, strengthen the light and shadow, weaken some 3D, characters are cute, sensitive to painting style, painter string; painting style compatibility is good, some painting styles will be a little abstract.
V20版本：回归到v18 v17的风格，增强质感，风格化。
Continue v18.v17 style, enhance texture stylization.
V19版本：奶奶比上个版本大，算不算优点，^_^更加明亮鲜艳，线条更加细腻，较上个版本减少油腻感和风格化，在2D和3D中找到平衡。
The breasts are larger than the previous version. Is this a merit? ^_^ They are brighter and more vivid, with more delicate lines. Compared to the previous version, the greasiness and stylization have been reduced, Finding balance between 2D and 3D.
18版本 强化了手和脚，还有略微调整，tag精准，很高兴我在熟女和萝莉之间找到了平衡，它们都可以完美实现，图像偏柔和一些，细节比较多。
Version 18 enhanced hands and feet, there are slight adjustments, tag accuracy, I am glad I found a balance between mature women and loli, they can be perfectly achieved, the image is softer, more details.
17 例图没有添加任何lora，更加完美的手脚！！！颜色重新调整了，使其更加鲜艳！比14版本弱化了3D。更加适合新手，简单的tag可以制作出高质量的图像。
Example picture without adding any lora, more perfect hands and feet!!! The color has been readjusted to make it more vivid! Weakened 3D than the 14 version. More suitable for beginners, simple tags can make high-quality images.
16 优化了直出效果，比前一个版本增强了3D，不想3D可以下载15.在不添加大量质量词的情况下也能很好的出图。同时提升了对御姐的理解，可以跑出大约25~28岁的成熟女性，强化了肢体的稳定性，其他轻微调整，使我感觉更舒服。（非常抱歉开通了EA，我需要Buzz支付每周的竞价和在线绘画，还有后续的计划。）
16 Optimized the direct output effect, which enhances 3D compared with the previous version. If you don't want 3D, you can download 15. It can also draw pictures well without adding a lot of quality words. At the same time, it improves your understanding of the imperial sister, and can run out of mature women aged about 25-28. It strengthens the stability of the limbs. Other minor adjustments make me feel more comfortable. (I'm very sorry for opening EA, I need Buzz to pay for the weekly bidding and online painting, and follow-up plans.)
You can use ADetailer to enhance facial details：https://github.com/Bing-su/adetailer
可以使用 ADetailer 来增强面部细节：https://github.com/Bing-su/adetailer
推荐我喜欢的分辨率：768X1344 832X1216 768X1280 704X1408 1024X1536
Model Recommended Resolution：768X1344 832X1216 768X1280 704X1408 1024X1536
Illustrious 1.0（noobAI）：896 x 1152、1216 x 832、832 x 1216、1344 x 768、768 x 1344、1536 x 640、640 x 1536，
Illustrious 2.0{Rouwei Illustrious1.1}支持更大分辨率：1536x1536、1728x1344、1344x1728、1824x1248、1248x1824、2016x1152、1152x2016
正面词 Positive Prompt：
masterpiece, best quality, amazing quality, very awa,absurdres,newest,very aesthetic,depth of field, highres,反义词 Negative Prompt：
worst quality,normal quality,anatomical nonsense,bad anatomy,interlocked fingers,extra fingers,watermark,simple background,transparent,low quality,logo,text,signature,face backlighting,backlighting,Recommended settings:
Steps: 25-35
CFG scale: 3-6
Sampler: Euler a Euler
Description
Continue v18.v17 style, enhance texture stylization.
回归到v18 v17的风格，增强质感，风格化。
FAQ
Comments (35)
V19 may be the first version that I like more than V13. The ones between kinda did not hit the mark for me. The only minor issue is that it tends to generate black-and-white images more often than the previous releases.
Good job and looking forward to any updates!
I think 19 the worst one, for nsfw... works very bad with promts
Hello everyone, how to know each use of these tags "masterpiece,best quality,amazing quality,very aesthetic,absurdres,newest,extreme aesthetic,in the style of fashion photography,light particles,cinematic lighting,Visual impact,sharp focus,Emotionalization,depth of field,dutch angle,scenery"
Thank you.
honestly, I would try each of those tags in a baseline generation to check the effectiveness of each tag... I personally don't use other quality tags other than masterpiece, best quality, absurdres, and highres . The rest is either atmospheric addition (light particles) or camerawork (dutch angle)
So yeah, maybe try adding them one by one on a baseline generation, then put them in combination and see how it impacts the output
Iam still using v18 still, seeing you are continuing on it with next version is great. Thx for a very good model and providing it for free :)
i enjoy v19 the most but 20 is cool too! one obsession is my favourite model!
this model is goat!!!!
这款模型是史上最棒的！！！！
请问大佬，mature版本是Vpred训练的，这个版本目前用的是EPS还是VPpred？
Having a weird thing happen when adding hairband or red hairband tags to images. It creates a gothic horror-esque scene
I'm actually experience a lot of issues of either horror like generations or no generation at all with just simple accessories that are tags and recognized. Kinda weird with v20 cause v17 doesn't have this issue.
I tried all the versions and I think 18 is the best, less artifacts, more natural background, more details, also unique characters from simple prompts
V20和19太太太太太太太太太牛逼了大佬！！永远支持你
nice work
V20 finally tones down the shiny skin (a little bit). The realism is really strong, but without losing the expressiveness and variety like you see happen in the SUPER-fine-tuned-for-realism models. For anime coloring/shading styles, I prefer V16 and V10 still, franky, despite the substantial improvements to prompt adherence since then. But V20 is ... headed back in the right direction after V19. V18 is probably still my favorite overall but it's super close.
V19和V20在webui上跑到95%后就tensor nan，在comfy上则能顺利跑通，是平台的兼容性还是？
webui不建议用
去更换reforge webui，sdxl基本都能跑
v20画黑皮的能力很弱
I've been away of the AI image generation scene for some time, so I'm intrigued as to how people animate their images such as you've done with yours in this latest model iteration. I'd very much like to try the model and use animation to do this, and see where I can get to. Would someone mind telling me if there are tutorials here on Civitai regarding that, or what should I google for in order to find something related?. Thanks! looking for forward to work with your model.
WAN 2.2 I2V is what you need. I2V = image to video. Look for tutorials for installation and usage locally on your PC to have more freedom. You can make 5 seconds videos or longer depending on your hardware.
@k1ngfloyd You're amazing, thank you!
@k1ngfloyd thank you
非常好，光影效果要比wai的版本更好
v20与之前的版本比起来，提示词的写法有什么变更吗？我使用了一些在v17和v19上能正常生效的提示词，不改变其他设置，只更换模型为v20，结果无法正常生成图像，主要是多人提示词，只会输出一个角色而且很模糊。另一个就是v19还在正常使用的画师串，同样的不更改其他设置，切换至v20后大概率会花图，跑5张图可能就花了34张。
this happens to me too
My opinion is that V18 reads the prompt best (extremely accurate) and the result is semi-realistic. V19 and V20 are not as accurate on the prompt, but they offer more flexibility on different styles.
❤️
a creating model for support furry and dragons?
Really nice model. It adapts well to prompts and the image distortions is minimal ✅✅
Even in my first attempts, I'm getting pretty good results 😸 This could be one of my next main models.
牛逼了哥们，生图界最锋利的剑
V20 is very senstive. One "wrong" prompt (not working, perfectly works with other model / previous version) and your picture is gone. 19 is much better but the best still is the V18.
为什么mature模型比bold模型细节好多了，还细腻多了，但是只能画成女？？
你猜为啥只能画成女（手动狗头
不知道为啥 眼睛画的总是容易很怪
I know this is a dumb solution but I regularly hires + face fix again after upscaling, that fixes the eyes 90% of the time. This is my favourite model on Civitai but I don't have enough buzz to be the only one auctioning it :(
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.