If you like my work, drop a 5 review and hit the heart icon. it's free and keeps me motivated
this is a new illustrious merge. still trying to know this base model more. It might have bugs and is definitely not perfect.
Free GPU use on Tensor
Mage provides Unlimited Generations
Patreon account. do check it out
I post my upcoming concepts and nsfw lora early on my patreon
If you want commission dm me on Discord.
Get early access to my upcoming NSFW Lora in my Patreon .
Support my work by joining any one of them and get early access to all my upcoming loras and other perks such as fan requests and Discord role.
Suggested settings:
I had CLIP skip 1 on every image
Samplers : Eular A, DPM++ 2M
CFG : 5-6
I had ENSD: 31337 all of them
All of them had highres.fix or img2img at higher resolution.
I don't use restore faces
I use afterdetailer for face details
Tiled diffusion for img2img upscaling and sometimes Noise Inversion for some more extra details.
4x-Ultrasharp upscaler
positive: masterpiece,best quality,absurdres, amazing quality
negative: bad quality,worst quality,worst detail,sketch,censored,watermark, signature, artist name
Join my Discord Server
Description
I had CLIP skip 1 on every image
Samplers : Eular A, DPM++ 2M
CFG : 5-6
I had ENSD: 31337 all of them
All of them had highres.fix or img2img at higher resolution.
I don't use restore faces
I use afterdetailer for face details
Tiled diffusion for img2img upscaling and sometimes Noise Inversion for some more extra details.
4x-Ultrasharp upscaler
positive: masterpiece,best quality,absurdres, amazing quality
negative: bad quality,worst quality,worst detail,sketch,censored,watermark, signature, artist name
FAQ
Comments (118)
Add this to civitai generator if possible!
not in my hand. lets hope civitai gives permission
@Goofy_Ai they are another indecisive bunch, allowing some but not allowing others
@ninjahatoori8 i asked them so hopefully if nothing gets wrong might get added to the system
@Goofy_Ai It's up there, but the selection of samplers is somewhat limited, so wondering how this model handles Karras? Or if I'm better off just sticking with Euler A? (Honestly, kinda rhetorical question, 'cause I'm just gonna try it and see how it turns out lol)
Was noobai or a merge of it used as base or merged in? Asking because would be a bummer otherwise since noobai has quite the extra knowledge
no.
noobai had trouble when merging. also loras were acting weird so just left it out.
I might try again in future with future noobai updates
Its hard to admit, but illustrious might just be better than pony
it definitely is - once you learned those booru tags, there's no going back anymore
@fhaifhai yeah i am having way more fun woth illustrious, the prompts are so responsive and i am always going to the danbooru search to find the proper tag spellings
@ninjahatoori8 my biggest hack lately was to install an auto-complete plugin in comfy that feeds off the danbooru tag list - doesn't get more convenient than that!
@fhaifhai i still have yet to setup stuff for local generation but i use tensor art and in tensorart they have a auto-complete feature, which is fiddly sometimes but it works
@fhaifhai wow , whats the name of the plugin??
@12ghjj i think it is comfyui custom scrips by pythongosss
@12ghjj yep, @Bipolaris was faster to respond. That is the one. You can use this very up-to-date list https://civitai.com/models/950325/danboorue621-tag-lists-incl-aliases-upd-1st-dec-2024 to feed into the auto-complete (just copy/paste)
Has pretty good prompt recognition and great detail/framing, but that ugly pseudo-realistic grainy pony style is impossible to avoid no matter how hard I try. Can't do official art/anime screencap images at all.
Same.
Agreed.
does this have a baked VAE or should i be using a specific one?
Baked in
prompts are super responsive and image consistency is really amazing, great job.
Best. As always.
amazing model
Great model, if someone knows how to describe two different characters in one prompt I'll be happy to know how
Generally just use their danbooru tags and nothing else will usually work out well
Great model. Where can I find a list of supported artist styles?`Some prompts seem to include them
I found this quite good: https://huggingface.co/datasets/Laxhar/noob-wiki/blob/main/danbooru_artist_webui.csv
@Bipolaris that is so so helpful, thank you
Edit: This model is more incredible than I initially thought! I spent a bunch of hours fine-tuning my standard keywords and the results blow the PrefectPony model out of the water! And there is much less manual editing necessary. This is truly glorious evolution!
As usual great work! This model seems very suitable for NSFW art. A small minus for me is that it defaults to a certain body type, which might not be an accurate representation of many characters
any tips? id like to hop on illustrious but i seem to be doing something wrong, images always come out decently overblown for me
The quality is surprisingly good even using TAESD VAE.
Is it normal for this to take significantly longer to render than the Prefect Pony XL model?
am i cooked or do pony loras work ? I tried running some for fun and they seemingly work? i thought pony loras dont work with illustrious
technically they would work as they both are sdxl based.
but because both of them are heavily fine-tuned the results are not 100%
Great model. Good quality,
This model has a huge potential, especially when loras starts working properly. The results even without them are already excellent. Maybe we've even finally reached Nirvana in our pursuit of lust
great model +10
Spectacular model
awsome model!
о боже, здесь столько ПОРНО!
谢谢
can you do a karin uzumaki lora too please? <3
Awesome
After I saw images from this model for prompt I was working, all I could say "what a fuck is this even!". JC this one is good, literally best images I seen so far.
My feedback so far:
+ Amazing default style
+ Lots extra styles and characters built-in
+ Prompt adherence is overall solid even for more specific things e.g 'hand in pants'
+ Works with pony Loras just fine
Downsides:
- By default seems very unstable handling duo prompts, worse than pony in this one regard.
- Very sticky watermarks in a large portion of the gens, especially when artist is specified.
If it wasn't for the weakness of duo gens this would probably the best model out there.
I'm still getting used to it since it's different from pony, so I might be doing things a bit wrong there.
Edit: For clarity with Duo gens I mean anything involving two specific characters esp with futa's involved. It seems to do perfectly fine for more 'vanilla' female x male scenes. Also compatibility with pony lora's seem a bit unreliable
What is a "duo gen"?
@Karsticles My guess is having more than one person being generated from the prompt.
@Karsticles Yes sorry with duo gen I mean any two character scenes, mostly nsfw sex positions and stuff. Pony seemed more flexible with those
@psyopsaiart thank you for explaining. I wonder what you are struggling to prompt?
@emmerich1227 thank you.
@Karsticles Any type of sex scene involving two people pretty much. After some more time spent illustrious type models though it seems to just requires a little different approach.
It still seems not as flexible as Pony in this one regard, but if you use regional prompter and other extensions with some clever prompting I think you can still get decent duo gens from this model.
prompt word hetero usually shows hetero, not futanari
@000111asder So this is something I wonder about: what impact does the continual addition of "clarifying" prompt terms have? Yes it will get more specific, but will it also increase the chance that my image fails to converge properly, or exclude other content that might have informed the image toward a better general image quality?
can you post a picture of your setting of your SD Interface? i cant find some of the setting you mentioned
Can you tell me the option you are having trouble?
Can't share so many options here at once
@Goofy_Ai the CLIP skip 1 on every image, I dont see a clip setting as well as ENSD, Highres.fix the dont use restore faces the tiled diffusion for img2img the 1x ultrasharp upsclaer. Im new to the program ive only followed a video to how to intall it
@Goofy_Ai maybe can you share your workflow :)
@ayyyylmaobby856 for SDXL type models and better you do not need CLIP skip set ever, as suggested in SwarmUI.
How to get free buzz hehe xd
Check out https://civitai.com/user/buzz-dashboard, and scroll down a bit
good
Is it based on noobai-xl?
I wasn't up-to-date with the SD community after the release of SDXL, and after returning to community after such a long time, THIS I find to be my new most favorite model in terms of RAW style.
Nice model, very easy to use due to having a nice 2.5d style by default. Vibe reminds me of sd 1.5, and anime models from those days
Super easy to use. With high quality images :D
any plans to make a vpred version?
Really nice model mate. Thumbs up!
I'm brand new to the AI scene, I've been trying out a lot of models, This is hands down the best one I've stumbled across. Thanks for your work!
I thought Illustrious was overhyped but I tried your model and it's amazing.
You always overdeliver 🔥
not that i care too much.. but it seems illustrious models are censored when it comes to poop content, just asking for a friend.
try "scat"
真的很好用
Good bot, do you have a list of trained character and their prompts?
What is the particularity of the "prefect" model compare with the "standard" model ?
Maybe a dumb question sorry, i am a Rookie
So very good model, but I have a tiny issue. The skirts are always between the legs, how can I make them normal, like floating normally?
Can't believe this worked last night and just doesn't work anymore (or as much) this morning... :///
read this guys https://civitai.com/articles/12904
its upto you guys to bid on it.
ask civitai not me
I'm not buying any buzz until this ones back for generation , thanks civ :)
why is not workin this model?
CivitAi thought it would be a good idea if people could only use some models if they could "bid" on them like in an auction. A weekly auction.
Most people think that's actually really stupid and they hate it.
@ChunkWit thx, indeed what a stupid idea
I keep coming back to this model, it produces amazing results and it's also not afraid of giving girls curves when you tell it too! Best model if you want exaggerated proportions without needing extra models that break things.
Any alternates for this?
With same features or even from same creator I missed this one...
It was making lights reflect and everything was realistic
@fardyfart0979 Nope, no other like this one.
So another week without this model. Make sure to bid on the auction page every week for this model.
@ImnotRyukoMatoi i found one its made by same creator and it has better lights etc.
It'scalled Prefect Illustrous XL
@fardyfart0979 @fardyfart0979 yes almost the same but not quite
Guys, make sure you bid on this model every week if you want it.
A week ago, this model won one of the 200 spots. I don't understand why it didn't win this week.
dude this model is insane, feels like u using a detailer without using it, i found it when its about to expire in hours damn, pls vote this lora
It's Really Nice!~
Great work!
This is my favorite model!
The aestethic and the detail it can achieve without using additional resources is so good.
So sad it lost the bid again 😔
whoaaa this model is online again, niceeeee
可
hi , i want to try this in "onsite only" generation in civitai , haha , no downloads needed
Когда будут обновления этой замечательной модели?
this is great!
thanks!
Great. What is the differense between your prefectIllustriousXL v10 and obliviousMix_v10?
This is amazing, thanks you for your work
Superb model!
Hope goofy_ai makes it active again, as its very consistent & makes the hottest pics
i do bid some buzz for it but civitai reduced the total checkpoint number for generation to 100. so unless users also bid double then its unlikely to return anytime soon
@Goofy_Ai Such an odd system. Thanks for explaining tho, & for putting these models together. Much obliged
@Goofy_Ai BTW can you tell me how to use your model while it is down? I cant select it but others are still genning with it, how are they doing this?
@partymantis they are using it locally on their pc
@Goofy_Ai Oh i see, interesting how the site still counts as your model despite folks doing it on their pc. Thanks again
@partymantis not really they are just manually uploading image after generation and its just showing under the model posts.
auction system is inherently a bad idea, though what else can one do? CivitAI needs money
this is genuinely a perfect model, i want to generate images without running locally but civitai kept taking it down...
i'm using a mac, which is impossible to run comfy fast. are there any methods of generating through cloud other than through civitai?
the owner actually fighting his/her way in auction with this model, so expect it not to be available on-site
You can rent an online server and do anything on it. Though ofcourse, it'll be expensive
google colab
i adore you
This is my favorite model. Will it be updated?
There's no better base model exist than this.
Details
Files
prefectiousXLNSFW_v10.safetensors
Mirrors
prefectiousXLNSFW_v10.safetensors
prefectiousXLNSFW_v10.safetensors
prefectiousXLNSFW_v10.safetensors
proto1_v10.safetensors
prefectiousXLNSFW_v10.safetensors
prefectiousXLNSFW_v10.safetensors
prefectiousXLNSFW_v10.safetensors
PrefectiousXLNSFW_v10.safetensors
prefectiousXLNSFW_v10.safetensors
prefectiousXLNSFW_v10.safetensors
prefectiousXLNSFW_v10.safetensors
prefectious_nsfw.fp16.safetensors
IL-prefectiousXLNSFW_v10.safetensors
prefectiousXLNSFW_v10.safetensors
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.