[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/t/ - Torrents

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • There are 82 posters in this thread.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


Janitor applications are now closed. Thank you to everyone who applied!


[Advertise on 4chan]


File: .png (18 KB, 399x433)
18 KB
18 KB PNG
NovelAI has been leaked and people need to have access to it! This AI will help change the world of digital art and smut forever. Requesting all seeders, help us out /t/eam C;

>NovelAI Torrent (pls seed when done):
magnet:?xt=urn:btih:5bde442da86265b670a3e5ea3163afad2c6f8ecc&dn=novelaileak

Pic related is all you need to download from the magnet above.

Stable diffusion tutorial: https://rentry.org/voldy

Skip step 3 and 4 entirely

What to do with the NovelAI files:

1. Rename model.ckpt inside animefull-final-pruned from NAI to final-pruned.ckpt
2. Place final-pruned ckpt and animevae.pt from the root in stableckpt in stable-diffusion-webui\models\Stable-diffusion
3. Rename animevae.pt to final-pruned.vae.pt
4. Make a stable-diffusion-webui\models\hypernetworks folder and place all the .pts from stableckpt\modules\modules from NAI in there

Total disk space used: 11gb

Settings -> Stable Diffusion : Set the model & module and click apply changes

Negative Prompt: lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry

Embeddings: https://rentry.org/embeddings
>>
>>1176863
Downloading with literally zero intent to seed, thanks OP!
>>
File: 172vu.png (714 KB, 1068x1478)
714 KB
714 KB PNG
>>1176863
Has anyone actually verified that the files are safe though? I've even seen it recommended that those who aren't familiar with python and pickle files wait for other people to confirm if it's safe first, since pickles are Python bytecode that apparently get executed when unpacking?
>>
>>1176879
I was about to comment the same, I'm downloading and seeding but I won't execute shit until I know it's safe.
>>
>>1176863
Been seeding since about 12 hours ago.
>>
> Total disk space used: 11gb

it's 55GB.
>>
>NovelAI Torrent (pls seed when done):
magnet:?xt=urn:btih:5bde442da86265b670a3e5ea3163afad2c6f8ecc&dn=novelaileak
there are no seeds right?
I only see 52 peers
>>
>>1176996
There are seeders but th torrent is fucked somehow and won't start on seedbox
Was able to download it using uTorrent tho
>>
>>1176863
>This AI will help change the world of digital art and smut forever.
kys coomer
>>
>>1176863
300gb uploaded, I've done my part
>>
File: file.png (550 KB, 512x640)
550 KB
550 KB PNG
>>1176879
>>1176880
it's safe
I generated pic related with it
the future of ai image generation is bright
>>
>>1176863
wait wtf is this?
>>
>>1177091
It's mostly for weeb coomers. For example if you have a favorite artist but he doesn't draw porn you can make the ai do lewd stuff inspired on his style.

This opens a huge can of worms because basically this ai does a perfect job as a forger(as long as you feed it with the correct references) and artists who live off commissions would be having a hard time getting work.
>>
Sorry for being retarded. I installed it but I can't find where you set the model in the webUI.

I know it's in the "Settings" tab but I can't find it for the life of me.
At the moment the images are pretty bad
>>
>>1176863

>1. Rename model.ckpt inside animefull-final-pruned from NAI to final-pruned.ckpt

rename modle.ckpt from NAI to final-pruned.ckpt , the fuck it have to names NAI

i have 32bit windows can i even play this ?
>>
don't know how to use it and probably won't bother, but i've added it to my seedbox :)
>>
>>1176863
>NovelAI
Legendary. Thank you. Seedin'
>>
>>1176863
It doesn't work for me. Can't initialize metadata
>>
>>1177145
wait a minute is this AI the source of the image in your last reply/ Because i've seen a few photos like that floating around the last week and I was wondering where the hell they came from. But this sounds cool as hell. God bless all these new AI and the techniques behind them. Software really has gotten insane in the last 5 to 10 years.
>>
>>1177145
wait a minute is this AI the source of the image in your last reply/ Because i've seen a few photos like that floating around the last week and I was wondering where the hell they came from. This sounds dope as hell. God bless all these new AI and the techniques behind them. Software really has gotten insane in the last 5 to 10 years
>>
Does this smut stories or images and is a gtx 1080 enough to use it?
>>
>>1177164
It's under "Stable Diffusion - Stable Diffusion Checkpoint".
>>
>>1177166
read https://rentry.org/voldy, you'll find out about lower than 4GB gpu memory guide, if that applies to you
>>1177312
I have a GTX 1080, and I run it fine on linux
>>1176957
biggest attentionspan on 4chan
>>
>>1177553
*I don't mean to say that you need linux, just that I am using the linux method of running it
>>
>>1176863
Linux users, when following the tutorial you may also have issues with activating venv; what you have to do is go to the stable-diffusion-webui directory and use 'python3 -m venv venv' and that will fix it.
I'll post up solutions to any more problems that I might run into, which I hope I will not have to.
>>
This is for image only, no text?
>>
>>1177553
dose anyone have windows 8.1 with activation key . i can't make python 3.10.6 run without it
>>
>>1176863
Thanks, I will seed it for 3 years or more. I don't know. I don't care. It's just 52 GB, so I'll seed it until the end of times.
>SNEED

>>1177014
And also thanks to you. I wanted to know if it was safe before opening or excecuting anything there.
>>
now the real question here, has anyone tried generating porn with it yet?
>>
File: 3.png (421 KB, 640x640)
421 KB
421 KB PNG
>>1177697
Yes.
>>
>>1177145
How can I find more of this kind of art? I've seen some posts on twitter and I was wondering if there are more?
>>
>>1177715
Go to any /sdg/ thread on
>>>/g/

There is an anon who posts AI drawings of those thicc women with that style.
>>
the guide says it may take 10-15 minutes tops for the first time run on step 7 and that it's "not stuck"., my shit has been running for an hour and a half and it just says "Installing torch and torch vision". wtf am I supposed to do now
>>
>>1177717
Read throught the script and figure it out.
>>
>>1177723
nvm it was just talking its sweet damn time. "10-15 minutes" my ass
>>
Wait so you just type in prompts into an AI picture generator and it will generate the waifu of your dreams in any outfit or situation?
Will it generate already existing characters?
>>
>>1177745
yes and yes, but it's finicky, takes a lot of trial and error
>>
Will this work on Mac? Are the embeddings required?
>>
>>1176863
I'm too stupid to follow these directions. How do I actually launch it after doing step 4?
>>
What does "Open webui-user.bat" mean?
>>
>>1177715
Just type "waifu diffusion thick" on google
>>
File: 1337797672274.jpg (6 KB, 210x173)
6 KB
6 KB JPG
For those wondering, If you dont already have a setup and are experienced with diffusion and running it locally through gitbash and python, this is 50gb of useless garbage. Also Nvidia only. Also, requires setup not mentioned here or in any of the guides. If your someone psyched about AI generation but its new to you, go anywhere else.
>>
>>1177828
>Also Nvidia only
I'm still looking into how to get stable diffusion running on AMD without a container, but this docker image works just fine out of the box:
https://github.com/AshleyYakeley/stable-diffusion-rocm
>>
>>1177828
you don't need to be experienced with this stuff, most people on 4chan use computers enough to figure out what they need to by googling. This was made as easy as you could get with this sorta thing
>>1177764
Okay, maybe I was wrong about the average 4chan user
>>
>>1176863
poorfag with a free vpn here, any anons think i would get in trouble if i started seeding openly?
>>
newfag to tech and github, do you have to re-download stable diffusion webui again just to have hypernetworks?
>>
>>1177828
I don't know python and I use Windows and I was able to set it up using a guide I've found on /g/
It's really not that hard
>>
>>1177887
right click INSIDE the folder, "git bash here", type "git pull" and enter
>>
Anyone able to help me out? Keep getting an error that torch can't use my GPU, tried the medvram/lowvram and python 3.7 fixes in the troubleshooting but no dice. Using a Ryzen 2600 4gb
>>
how does one with no knowledge of computers get into this stuff?
>>
>>1177828
This would've been good to know beforehand.
>>
Are there any other generators that will work with ATI GPUs? And preferably ones which are easier-to-use?
>>
>>1177928
*AMD
>>
>>1176863
>Skip step 3 and 4 entirely
then why make step 3 and 4? are u retarded
>>
>>1177943
are you?
>>
i've been stucked at " installing torch and torchvision " part and waiting like more than 1 hour anyone else got stucked here ?
>>
>>1177920
You give up.
>>1177962
Have you even read anyones posts here?
>>
>>1177943
step 3 and 4 in the tutorial you bimbo
>>
File: file.png (11 KB, 1406x36)
11 KB
11 KB PNG
>>1176863
Have a radeon GPU, so no stable diffusion for me
Still ain't stopping me from seeding the fuck out of it
>>
>>1178111
Do you mind dropping the magnet for that part2? I'd like to at least have a peek at whats in it
>>
>>1177828
>50gb
>Pic related is all you need to download from the magnet above.
>nvidia only
guide literally points to https://rentry.org/ayymd-stable-diffustion-v1_4-guide for amd
>Also, requires setup not mentioned here or in any of the guides
??? works on my pc
>>
File: 1665523215239018.jpg (36 KB, 474x502)
36 KB
36 KB JPG
>>1176863
>minimum requirements
>16 gb ram
I guess it was not meant to be
>>
>>1177828
I want to learn this. Do you have any guides or recommended sites?
>>
>>1178253
im doing it on a gtx 1660 with 6gb vram it works just fine
>>
>>1177831
Also Linux only
>>
>>1178289
Currently trying to use it on a 1650 super but it keeps spitting out black screens when I try to generate images, did that happen to you and if so did you find a workaround?
>>
>>1177711
>seven fingers
lawl
>>
>>1178116
magnet:?xt=urn:btih:a20087e7807f28476dd7b0b2e0174981709d89cd&dn=novelaileakpt2&tr=udp%3a%2f%2ftracker.openbittorrent.com%3a6969%2fannounce&tr=http%3a%2f%2ftracker.openbittorrent.com%3a80%2fannounce&tr=https%3a%2f%2ftracker.nanoha.org%3a443%2fannounce
>>
>>1178417
What's the difference in this one? Why is it double the size?
>>
>>1178317
Yes, but if you read the thread at all you'd know there's a Windows solution already. There's no excuse, anon.
>>1178148
>>
>>1178339
the workaround is mentioned in OPs guide. that's a common error in 16xx series cards due to low vram you have to run on full precision and medium vram
>>
>>1178463
According to the readme file
>This LEAK contains NovelAI's "production models" (whatever those are)
>and some other random Stable Diffusion models.
>>
File: 1666064846139.png (487 KB, 512x768)
487 KB
487 KB PNG
its funny what this thing can and cannot draw. a woman with a buzzcut is incomprehensible to it but it can do abs like a renaissance painter.
>>
File: 1666064890643.png (381 KB, 512x768)
381 KB
381 KB PNG
its better a doing abs than tits, honestly.
>>
For those curious, I found a solution for AMD, which I've tested on Linux and I'd like some feedback from Windows anons. It all has to do with python so I don't see why it wouldn't work for Windows.
I used this guy's instructions at first then tried without the docker container with success:
https://mcgillij.dev/stable-diffusion-on-amd-cards-with-rocm-on-arch-linux.html

So basically just follow the regular stable diffusion setup steps, all the way through conda and whatever until you try to run it and it crashes complaining about Nvidia.

Then it's just a matter of updating your python packages:
conda activate ldm
pip3 install --upgrade torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/rocm5.1.1

tl:dr if you're on AMD run the pip command above
>>
is art dead? Ive spent like a couple hundred hours over the past 2 years learning how to draw. I guess I should learn how to animate then?
>>
>>1179624
it isn't it takes a lot more time to get the perfect image of your desire by ai generation. clients usually want small details and if you are in a company things have to be very specific to the game designers requirements. Yes you can use AI to get a base template and honestly that seems like all it will be good for. AI won't be able to replace a professional artist. So keep practicing, keep drawing. No one is stealing your job but your mindset. Use AI to save time or get ideas to fill a blank canvas.
>>
>>1179624
>animating
https://github.com/HelixNGC7293/DeforumStableDiffusionLocal
+ some new releases supposedly coming next week
>3d modeling
https://github.com/vt-vl-lab/3d-photo-inpainting
https://airender.gumroad.com/l/ai-render
>text writing
story gen is already way further than image generating
>programming
github copilot

nothing is safe anon. every field is getting attacked at the same time
but it's not really a problem? if you're technology literate all this does is give you huge advantages over others
just stay up to date on it (https://rentry.org/sdupdates is great), think of ways to add it to your workflow and meanwhile keep doing whatever you want
>>
anyone know how to restore web ui settings to default? I made a bunch of changes testing something and its kinda fucked rn
>>
File: 00096.png (428 KB, 512x512)
428 KB
428 KB PNG
>>1179745
Just a stretch but in the webui directory, try "git status" and see what files are modified. You can then use "git restore" on any files that look like config files. The config may be untracked but it's worth a shot.
>>
File: file.png (40 KB, 980x292)
40 KB
40 KB PNG
hello sers i'm doing the needful

>>1179745
delete config.json and ui-config.json
>>
Seeing a lot of results with nudity (nipples generating, vagina, etc) but I can't seem to get that. Any step I'm missing, or is that part of the photo edited in?
>>
>>1179832
No it's not edited, this is straight out of the AI. Use "nude" or similar as one of the prompts.

As a side note, I noticed that some releases do a safety check and censor nsfw photos but the webui from the guide has a checkbox in settings that should be diabled by default.
>>
>>1179843
Nsfw results are on and everything, just doens't want to generate anything. Pic related, is what I got from an nsfw tag.
>>
>>1179845
>>1179843
Nvm, am just retarded
>>
Sorry, where do I find def prepare_environment to run on CPU?
>>
>>1177828
i literally only download this to seed. generating the pics will come later, when i get enough GPU power
>>
>>1178263
look at /sdg/ on >>>/g/
>>
How do i load my modules or do they auto load
>>
>>1176863
Does any of this NovelAI leak come with writing stuff? If so, how do I put it to use? No guide for written material.
>>
What ESRGAN model should I use in order to clean up and upscale the anime images and keep it the cartoon style, and which should I use in order to make it look more realistic? What do you guys use?

I'm looking at this https://upscale.wiki/wiki/Model_Database
>>
>>1179828
Where is pt2?
>>
>>1180458
Optimal method is to fuck around and find our. I played a fair bit with a lot of the models there and found that despite most pics tested were "anime" the bet results were rarely obtained with the same model.
>>
bump
>>
I heard there's a part 3 to the leak now?
>>
>>1176863
Does this novel ai thing that you are saying to seed only make waifus and nsfw stuff or can it also make ai generated images
>>
>>1185912
the waifu's are AI generated. You can also add nsfw to negative prompts to blacklist nsfw. The novel ai model seems to generate anime-ish stuff mostly since it was trained with that mostly. Hope that answers your question
>>
bump
>>
>>1179828
>>1185156
I'm gonna need to know where to find these. For science, of course.
>>
Is this what websites like deepsubeke.io use to nudify irl women?

What software lets you take pictures of women and nude them/what are the requirements from a hardware standpoint?
>>
So apparently my gpu is too old to use the normal version of pytorch so I have to build my own apparently. Where would I put my custom pytorch build when it is done?
>>
Threads on /g/: https://boards.4channel.org/g/catalog#s=Stable%20Diffusion
Guide: https://rentry.org/voldy
News: https://rentry.org/sdupdates3
A curated collection of up to date links and information: https://rentry.org/sdgoldmine
>>
>>1190026
thanks a lot for this brother
>>
>>1179102
First bit of feedback: what exactly are we supposed to do if we're on Windows? The guide is for Arch, so the only thing I can think of is that we're supposed to use chocolatey instead?

What does he mean by adding ourselves to the docker group?

The AMD/Windows guide in nai-speedrun doesn't help much either - doesn't say what to do to actually run the master test. Do we have to do something with Docker?

Who do I have to pay to get a list of clear steps to follow in order to get the leak working on a Windows machine with an AMD card?
>>
>>1191054
>What does he mean by adding ourselves to the docker group?
It's a Linux thing, there's a group named docker that has certain permissions. Not really relevant if you're not using docker. Docker shouldn't even be necessary, there are just some docker images out there that make things easier since they have the right python environment and stuff setup.
>The guide is for Arch
Basically ignore the Linux specific commands and just do the Python related stuff. The pytorch stuff is all that matters, really.

Basically, Stable Diffusion was made to run on Nvidia hardware, but we can do extra steps to fool the system into thinking our AMD card is an Nvidia card.
>>
has anyone figured out how to do this on invokeai rather than automatic1111? i am trying to run this on my mac
>>
bump
>>
>>1191519
Yeah, just don't use a mac idiot
>>
There's a Stable Diffusion thread: >>1190826
>>
Is it possible to generate text with the novelai leak?
>>
File: pjNQRPQ.png (261 KB, 512x512)
261 KB
261 KB PNG
I was able to get everything going up until the Asuka test where I got this image and found this quote in the troubleshooting guide:

are you a cpu toaster bro? i don't know why, but running webui in cpu-only mode yields completely different results. (btw my new-ish laptop ran this at 15s/it, i'm praying for you...)"

Specs:

CPU: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz 4.00 GHz
GPU: NVIDIA GeForce GTX 970

I have 16 gb of RAM and 4 bg of vram and made sure to add --medvram to the webui-user.bat

wtf do I do
>>
Bump
>>
Well isn't this neet.
>>
>>1197272
to this...
>>
>>1179692
what can i use for text generation?
>>
File: screenMem.jpg (328 KB, 1200x900)
328 KB
328 KB JPG
>>1176863
Now all we need is a text one from c.ai or novel to leak or weights for palm/lamda.
>>
I'm about to download this to all 3 of my seedboxes and host it for you guys. It should download a lot faster once I'm done. Hopefully around 50-100mbps
>>
>>1178417
What the fuck is in part 2?
>>
>>1176997
>seedbox
It doesn't work on UltraSeedBox, but it works on HostedByDesign.
>>
>>1201711
probably more refined and improved data sets which lead to better results



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.