If you want to use more my checkpoint online generation, please visit here.
https://tensor.art/u/762555264535746522
WAI-NSFW-illustrious-SDXL - SeaArt AI Model
Character Select SAA
This is a Stand Alone App with AI prompt, Semi-auto Tag Complete and ComfyUI/WebUI API support.
Now supports 5058 (includes multiple costumes) Character list.
https://github.com/mirabarukaso/character_select_stand_alone_app
One-Click embedded package
In case you don't know how to build your own Python enverment, try the embeded_env_for_SAA
Down and unzip to your computer
Db-click #run_XX.bat
https://huggingface.co/datasets/flagrantia/character_select_stand_alone_app/resolve/main/embeded_env_for_SAA.zip
下载解压到任意目录,运行 #run_CN
Download and extract to any directory. Run #run_CN or #run_EN
Online Quick search character
Character Select Saa - a Hugging Face Space by flagrantia
Please do not add too many quality and aesthetic-related tags, nor overly long negative prompts, as this will actually reduce image quality and make it more blurry.
v17
Fixed issues where some characters had inconsistent color tones with the background.
Attempted to fix the issue where the background had low relevance to the characters.
The overall coloring in this version is slightly smoother and cleaner than before.
Attempted to improve the high-resolution fix’s ability to correct limbs; using it has a high chance of automatically fixing issues with arms, legs, and hands/feet.
If you're not satisfied with the default character age, try adding (aged up:1.0–2.0) or (mature female:1.0–2.0) at the beginning and adjust as needed.
ps
Regarding model updates: I create models purely as a hobby, not as my main job. So I work on them in my spare time, and the update frequency depends on my availability. Recently, my computer suffered hardware damage—both the GPU and hard drive were affected. As a result, updates will likely be slower for quite some time. Thank you for your understanding.
Recommended software: forge-neo.
Recommended settings:
Steps: 15-30
CFG scale: 5-7
Sampler: Euler a
The VAE is already integrated, please do not ask such questions anymore.
use size larger than1024x1024 for the original dimensions.
Example images use 1024×1344.Hires upscale: 1.5, Hires steps: 20, Hires upscaler: R-ESRGAN 4x+ Anime6B,Denoising strength: 0.35~0.5
There are four safety rating tags: general, sensitive, nsfw,explicit.
Users are expected to consciously add "nsfw" to negative prompts to filter inappropriate content.
Positive Prompt
,masterpiece,best quality,amazing quality, Negative Prompt
bad quality,worst quality,worst detail,sketch,censor,V5-V15 User Manual
V1-V4 are outdated,please do not download them anymore.
This is a list of characters that have been tested and are implementable on V11.
sieecc/WAI-NSFW-illustrious-SDXL at main
These are just temporary and not all tests have been completed yet; more will be added subsequently.
Recommended settings:
Steps: 25-40
CFG scale: 5-7
Sampler: Euler a
The VAE is already integrated, please do not ask such questions anymore.
use size larger than1024x1024 for the original dimensions.
Hires upscale: 1.5, Hires steps: 20, Hires upscaler: R-ESRGAN 4x+ Anime6B,Denoising strength: 0.35~0.5
There are four safety rating tags: general, sensitive, nsfw,explicit.
Users are expected to consciously add "nsfw" to negative prompts to filter inappropriate content.
Positive Prompt
masterpiece,best quality,amazing quality,Negative Prompt
bad quality,worst quality,worst detail,sketch,censor,Description
FAQ
Comments (112)
Love it so much.
BABE WAKE UP NEW WAI ANI ILLUST MODEL JUST DROPPED
let's gooooo!!
O.O
Oh my gotto a new WAI version, thank you lord!!!
works like a charm
be my eyes deceiving me? new update from totally BASED WAI just dropped?
My favorite model, thanks for the new version <3
我很喜欢你的模型和你的作品,非常感谢。
pony loraも一部はponyで使う以上に機能にするのが素晴らしい。凄く機能する物と、全く機能しない物の違いは不明ですが
is it better using quality prompts only 'masterpiece, best quality, amazing quality'
or using more just like 'very aesthetic, absurdres, newest' ?
'very aesthetic, absurdres, newest'
These will change the fundamental art style, please add according to your own situation.
Nice and all. But all Models done before the 8th version are now unusable on civitai, because there is no option to use the old version. Sure you can use the new version with the models made with earlier, but the look awful now.
Please make the older versions available on civitai.
It's not that I don't want to do this, it's civitai doesn't allow it.
@WAI0731 I know :( civitai's "don't look back" policy is kinda telling
画人物背身的手臂非常准确,太强了
请问这个model有教程之类的吗?
Is there any tuturial for this model?
正负面一打 质量词一打 其他不就是你自己发挥 你说画师串的话,看说明文 有一些画师跟角色参考的链接
Other models are very hit or miss with Loras, but this model makes them look good so consistently. You must’ve used some magical spell to make this
训练集是不是有问题,练出的图经常出现书本杂志之类的小本子?
Great model. Works well overall. This model needs a bit more detailed prompt. Otherwise it produces simple results.
LORA compatibility is good.👍
It recognizes fictional characters but not completely.❓ It is hard to get good results just by typing the character's name.
Hands and faces are good but not always.❓ It may take some work to get good results. Fingers can be distorted, especially when holding a weapon or in slightly different poses.
It works at high resolutions but not great. Just not too bad.
I still think it can be improved.
The changes in this most recent version (version 8) feel like they've backfired slightly to me. I'm noticing far more phantom characters that weren't asked for in the prompt in this version, and it doesn't feel like there's been much improvement in other areas such as character recognition or more concepts. I'd rather still be using version 7
Wow! This checkpoint is amazing. I'm especially impressed by how well it handles multiple requested, very detailed characters engaged in physical interaction without mixing their characteristics. It also produces very accurate anatomy, including hands. And all of this with zero LoRA used. :o
It has one issue: it tends to place a character in the background, which can yield fun results like a character in bg catching the other characters having fun in fg. But it becomes weird when the character in the bg is already in the fg. Is there something I can add to the prompt to avoid that, or duplicates in general?
Checkpoint, which you created, works very well with the Lora's I created and I love using it.
Your regular updates are also greatly appreciated.
I would be very happy if you could continue to update at your own pace.
Thank you for everything👍.
if someone know "horsetail" artist, i cant seem to get the artist style
i want to do something like this but i dont know who and how he does it, theres literally no loras for it
https://www.pixiv.net/en/users/96418447
@peenusbutter i have put the tag "horsetail" but seems it's not working
Time to train your own lora on the images then !
@XTraitor im more of a consumer type so yeah xD
Weirdly with the v8.0 the breasts and ass have bias when you want to do anthro. it make anthro with furless (or scaleless) breasts and ass.
I love this model. Other models ive tried have a harder time recognizing characters and actions, which meant that i had to rely somewhat on Loras. This one hardly requires them. Great job.
Chair
Chair?
Yes Chair
Chair!
can you do a karin uzumaki lora too please? <3
Would it be possible to include more training on "dashed eyes" and "cross-eyed"? These two tags don't really work from my testing
Why 1024x1360? Shouldn't it be kept /64 like 1024x1344?
This is my personal preferred size, and there isn't any particular reason for it.
欧金金有时自带打码,怎么去掉🤣
@WAI0731 Where can I see the list of artists included in this checkpoint?
You can try using artist tags from Danbooru as a reference, but it's not guaranteed to work 100%.
@WAI0731 ok! I'll try. I also wanted to tell you one thing. The Illustrious checkpoint is outdated, it has no support. Don't you think it would be better to use the Noobai XL checkpoint as a base? Laxhar Lab continued to develop it. Now they have just released the full version V-Pred 1.0
@megachina31687 "Illustrious is outdated", "they continued to develop Illustrious" What? Your word is contradict each other, read again
@WAI0731 what following Tags did i need to use to activate their artist style art? Like "drawn by WAI0731"?
@Catnonimous just name
@Catnonimous Corrected
is this model can use e621 tags?
sexy ass gen, great for hentai ❤️
Incredible model!
it took me a while to get anything worthwhile out of this, probably because it's capable of so many anime styles, but when you get the hang of it, it's incredible
question are those pony lora compatible with this ckpt?
Hello, is there any SDXL version of WAI NSFW to use on civit?
It's very versatile
nice
Nice
Wow, that model is absolutely amazing :D No matter if SFW or an NSFW, it creates fantastic images!
Only downside: Sometimes, if you ask for NSFW stuff, censorbars are put in place. I was able to go around this by adding "censorship" and "censored" into the negative prompts, but hey, as long as the negative prompts work I am fine ;)
solid model
If I can speak Japanese on the text prompt, I will keep up with it for the rest of my life.
Good
GoodArt
Nice
ass
牛逼呀
good
Great checkpoint! I wish it could pull data from Gelbooru too, since Danbooru has some censorship and restrictions on banned artists.
I just want to know if this checkpoint was trained to work well with most hentai or anime characters. I'm trying to avoid using any LoRAs. Can anyone verify this for me? <3 Thank you! 😇
where do you put lllustrious on Forge?
models\Stable-diffusion
@WAI0731 interesting. i usually put my checkpoint in there. Okay dokie. Thanks for the info
Help noob, please. Why the images are so desaturated? Is it requires to install anything else?
reference v8 Recommended settings and not use vae
Could be that the cfg value is too low or you got an incompatible vae.
画福瑞都很nice,太强了
Good
cool
wow!!
Hello. Why do all my pictures turn out very bright, burnt and either pink or yellow when generated?
cfg too high or got an incompatible vae.
It's just doesn't work. When i trying to generate something it gives me empty images with artifacts. I generate in Automatic1111 with 1024x1024 resolution even trying the same promts as on other images. What i did do wrong. Other models such as Pony work with Loras correctly.
Pony lora may not always be compatible
@WAI0731 Thanks for answering i just reload UI and all seems start to working, i assume the problem is in promt that i copied from one of generations that using this model. When i paste promt it start generating a mess.
This a great checkpoint! In my opinion, version 7 has better body composition than version 8.
This is a great one to use
Is it based on NoobAI-XL?
No, but there's a model created by WAI0731 that's based in NoobAI-XL, that model is WAI-SHUFFLE-NOOB.
ni
Wow!
Good model.
It have a problem depicting non Human Characters
Same here.
Added to my backlog of models to try!
Best one I've found yet
This is a highly competitive model for making NSFW content
awesome style
Can I kindly ask for a clarification? If I use an Illustrious checkpoint, can I use Pony and other types of Lora? Thx!
pony and illust are both sdxl
porny used wayyyyy different settings compared to base sdxl and pony
Cool
May I ask, do we have a list of available characters that we can generate without lora, just like pony?
I don't know all the specific characters this model recognizes, but, generally speaking, Illustrious models recognize character tags from Danbooru. Go to the Danbooru website and search for the character you want. (Characters have green tags.) This is important -> Copy and paste the character tag and REMOVE the underscore, and just replace the underscore with a space. For example: hatsune_miku should be -> "hatsune miku" in your prompt.
Style and Character reference
https://civitai.com/articles/8977/epis-embedding-maps-and-graphs
@CHROME2D0ME That works for me, but I assume it only recognizes characters with large data on danbooru, yes?
@Kaze111 It should recognize characters with both large and small datasets. I would say it's less likely to recognize characters that aren't popular or very well-known.
pretty cool one
Awesome model! I can't decide which of the two models is better IL or Pony :)
No matter how hard I try or see shiny new illust I just get better results from pony. Probably because I’m just so familiar with how it works but I’ve seen absolutely amazing stuff out of illust. And doing score_9…. And the other 10 required intro tags doesn’t really bother me as I just have them saved on notepad. Seems to be the main reason people like IL is you need less words and manditoru tags.
Pony is a good model, but NoobAI-XL is better.
@CuauhtemocI5MAL can you explain why? Honestly would switch if I saw results
@greggoulet2870 Both models (NoobAI-XL and Pony) are responsive to prompts made with tags used in e621 and/or danbooru, but NoobAI understands more concepts than pony; NoobAI can natively replicate several artists' styles, while Pony can only natively replicate the style a tiny amount of artists; and NoobAI uses more conventional tags (masterpiece, best quality, etc.), while Pony uses score tags (score_9, score_8_up, score_7_up, etc.).
@CuauhtemocI5MAL at the upper level of quality and skill especially of nsfw it’s always pony I’ve seen it be done in hands down. So pretty much it’s just more user friendly is the argument. I literally can’t get an IL picture to look nice unless I directly rip from others with perfect ratios of artist levels and weird little Loras. Stuff like wai ani pony just shits out HD images 1MP that are mostly always great. Yeah you need to do score up but who cares if it just looks better.
cool
Could you please tell me which controlnet model can be used for this series model? xinsr or other
Cool
Details
Files
waiIllustriousSDXL_v80.safetensors
Mirrors
waiNSFWIllustrious_v80.safetensors
WAI-NSFW-illustrious-SDXL-v08.safetensors
waiNSFWIllustrious_v80.safetensors
waiNSFWIllustrious_v80.safetensors
waiNSFWIllustrious_v80.safetensors
waiNSFWIllustrious_v80.safetensors
waiNSFWIllustrious_v80.safetensors
waiNSFWIllustrious_v80.safetensors
waiNSFWIllustrious_v80.safetensors
waiNSFWIllustrious_v80.safetensors
WAI-NSFW-illustrious-SDXL-v08.safetensors
waiNSFWIllustrious_v80.safetensors
waiNSFWIllustrious_v80.safetensors
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.



















