stability diffusion vore

Keep our community informed! This forum is for discussing and sharing vore-related information. Post any relevant material and/or links here, and engage in conversations!
Forum rules
This is for general discussion, if you found something you want to post, please use one of the upload forum, if you made something and want to share them, please use the work to be shared forum!

Re: stability diffusion vore

Postby jimpickins » Mon Oct 31, 2022 10:29 am

I just use the automatic1111 repo. Probably the most feature rich out there and no NSFW filter
User avatar
jimpickins
Somewhat familiar
 
Posts: 112
Joined: Mon Nov 21, 2016 12:29 am

Re: stability diffusion vore

Postby jimpickins » Mon Oct 31, 2022 4:32 pm

I set up a stable diffusion vore server, PM me for the invite
User avatar
jimpickins
Somewhat familiar
 
Posts: 112
Joined: Mon Nov 21, 2016 12:29 am

Re: stability diffusion vore

Postby ryanshowseason3 » Mon Oct 31, 2022 8:24 pm

Rolled my own from the directions linked from 4chan. Was easier than I thought. It's been about 24 hours and these are the best internals I've generated. A lot of trash along the way but these are pretty usable with a few tweaks.
Attachments
tmpw17shc8a.png
tmpn4osu6jt.png
tmp0_xea6vc.png
tmp3doe40jd.png
tmpugpjjuph.png
tmpf5k4u8sv.png
tmp0kadpomd.png
tmph65vlxpb.png
User avatar
ryanshowseason3
???
 
Posts: 2479
Joined: Thu Jan 05, 2006 12:00 am

Re: stability diffusion vore

Postby masterofmanyforms » Sun Nov 13, 2022 11:28 pm

To report the hentai diffusion pack itself now recognizes the prompt vore belly but Novel still has no clue. I have started with tuning and using bodies as a base from crayon Ai. My results are starting to pop off. The red thing is the prompt incineroar with a vore belly. Image Funny how it got half right the other so wrong.

Regardless I have a bible of tricks using badly penciled sketches to get it to start and slowly shape the subject into what one wants. This also includes jumping between models. I lack the GPU required for any of the 3 big micro model training options. But I think with prep and the ability to train a dream or hyper custom variant the results could reach the level results this forum likely is looking for."
masterofmanyforms
New to the forum
 
Posts: 7
Joined: Wed Sep 01, 2021 2:08 pm

Re: stability diffusion vore

Postby fixated1 » Tue Nov 15, 2022 11:07 am

AI is the worst among many assaults against artists and people with eyes this year. It's only free for now. Once the public has done the developers' jobs of training it until it's finally producing good images it will become a yearly license costing thousands of dollars. Until then the only good it will do is get the "do you do requests?" People off our backs. It's the moral equivalent of "I'm not touching copyright! I'm not touching copyright!" And people have used it to copy artists less than 24 hours after they've died to get clout. I have zero trust in people to handle this with any sort of decency.
User avatar
fixated1
Advanced Vorarephile
 
Posts: 913
Joined: Sun Nov 22, 2015 1:38 pm

Re: stability diffusion vore

Postby AlluringPredation » Wed Nov 16, 2022 10:41 am

First thing about AI and tldr; AI can not recreate precious OCs with accuracy, it can not do complex characters, it can not do interaction. Artists still have a job now and for a long time in the future. Here's a professional artist weighing in on the discussion. https://www.youtube.com/watch?v=zc5EDf7CYKk&t=2562s

AI is great at general concepts, most of the time. You wanna see a cute cat girl with a belly? DONE! You want your specific character doing such and such? Or some unrepresented fetish? Not a chance. It can do a lot of your FOTM type characters since they're well represented in the database, but people's OCs are untouchable territory for the AI. It's terrible at persistent traits. If you have a more generic character you probably could train some embeddings, but it'd still look off in many instances.

Once the public has done the developers' jobs of training it until it's finally producing good images...

That's not how these AI models work. All people are doing if fine-tuning for a specific use, this does nothing to affect the base model. Do you think the creators of this software want their AI chock full of yiffing and vore? It's a common and understandable misconception, but nothing we do affects the base AI.

It's only free for now.
>...it will become a yearly license costing thousands of dollars.
It is open source and can be run locally with a NVIDIA GPU with as little as 2GB of VRAM. There coming versions will necessitate higher VRAM requirements, with the finalized versions likely needing a 3090/4090, so being that its able to be run on a consumer level GPU, doubt we're going to get the point you have to pay thousands. Especially since the base models are already out in the open for public use. Even then they're making optimization improvements and will likely reduce the future requirements, https://syncedreview.com/2022/11/09/alm ... ng-can-be/

And people have used it to copy artists less than 24 hours after they've died to get clout. I have zero trust in people to handle this with any sort of decency.
Yeah, people are asshole. Just like with anything it is up to the individually to use responsibly. And that isn't being done.
Discord for artists and those who commission them! https://discord.com/invite/F3UDqqAx53
User avatar
AlluringPredation
Participator
 
Posts: 293
Joined: Fri Mar 27, 2015 11:10 am

Re: stability diffusion vore

Postby ryanshowseason3 » Wed Nov 16, 2022 12:43 pm

fixated1 wrote:It's only free for now. Once the public has done the developers' jobs of training it until it's finally producing good images it will become a yearly license costing thousands of dollars.


Sorry but this a bad take with demonstrable evidence otherwise. Ignoring the rest of this post. There are lots of ethical and moral questions but this part is simply false.

Yes it's already been monetized. And no I'm still able to build new models despite that. It's open source. I've already cloned the repo.

There is no taking that back and you can create your own models. You've only demonstrated how little you know and your lack of research. Stick with the moral and ethical side, you're not wrong but maybe not in this thread? Seems maybe off topic when the question was "is it possible" and not attacking it from every angle you can think of.

The real ones under attack are stock image sites. Further demonstrated by one of them outright partnering with Dall e instead of fighting them.
User avatar
ryanshowseason3
???
 
Posts: 2479
Joined: Thu Jan 05, 2006 12:00 am

Re: stability diffusion vore

Postby fixated1 » Fri Nov 18, 2022 2:33 am

ryanshowseason3 wrote:
fixated1 wrote:It's only free for now. Once the public has done the developers' jobs of training it until it's finally producing good images it will become a yearly license costing thousands of dollars.


Sorry but this a bad take with demonstrable evidence otherwise. Ignoring the rest of this post. There are lots of ethical and moral questions but this part is simply false.


It's not, and history backs me up. There are plenty of programs that have been open source or had open source counterparts that have gone to licensed models. Especially art programs. This will absolutely go to payment plans and no amount of you downloading a janky beta will stop it. Trust me as someone who's been an artist for a while.
User avatar
fixated1
Advanced Vorarephile
 
Posts: 913
Joined: Sun Nov 22, 2015 1:38 pm

Re: stability diffusion vore

Postby ryanshowseason3 » Fri Nov 18, 2022 7:39 am

fixated1 wrote:
ryanshowseason3 wrote:
fixated1 wrote:It's only free for now. Once the public has done the developers' jobs of training it until it's finally producing good images it will become a yearly license costing thousands of dollars.


Sorry but this a bad take with demonstrable evidence otherwise. Ignoring the rest of this post. There are lots of ethical and moral questions but this part is simply false.


It's not, and history backs me up. There are plenty of programs that have been open source or had open source counterparts that have gone to licensed models. Especially art programs. This will absolutely go to payment plans and no amount of you downloading a janky beta will stop it. Trust me as someone who's been an artist for a while.


Lol what you said already happened. It went subscription based on a dozen sites. What you said hasn't happened. The cloned repositories didn't magically poof.

Guess time will tell in a year. No point in arguing further we'll both see where it ends up.
User avatar
ryanshowseason3
???
 
Posts: 2479
Joined: Thu Jan 05, 2006 12:00 am

Re: stability diffusion vore

Postby fixated1 » Fri Nov 18, 2022 12:43 pm

There will always be open source counterparts to varying degrees of quality. Maya's is Blender. Photoshop's is GIMP. Toon Boom's is Opentoonz, and so on. But whoever gets popular, streamlined, good, and (god forbid) industry standard, is going to bend you over a barrel for the privilege. Go look at individual licenses for Maya and Toon Boom. Pantone is charging artists $15 a month for their license... on colors. I want this unethical set of programs to wither and die, and I want to see the people claiming to be artists because they type words in the stealing machine to get their come uppance. But I know where this is heading, and at least they'll feel that pain in their wallet like I do.
User avatar
fixated1
Advanced Vorarephile
 
Posts: 913
Joined: Sun Nov 22, 2015 1:38 pm

Re: stability diffusion vore

Postby fanitfam » Wed Nov 23, 2022 6:13 pm

ryanshowseason3 wrote:Rolled my own from the directions linked from 4chan. Was easier than I thought. It's been about 24 hours and these are the best internals I've generated. A lot of trash along the way but these are pretty usable with a few tweaks.

These look very high quality, I can't seem to get anything even remotely near this on my own. Can you walk us through how you managed to get outputs like this either here or in Jim's SD discord? Did you get these through prompts alone or img2img? Did you use any inpainting to smooth out these results? What kind of tags did you use? What checkpoints/models did you use, did you use one of the original models in the voldy guide or did you also make use of the NovelAi model leak?
fanitfam
New to the forum
 
Posts: 1
Joined: Tue Nov 22, 2022 6:31 pm

Re: stability diffusion vore

Postby ryanshowseason3 » Sat Nov 26, 2022 5:27 pm

fanitfam wrote:
ryanshowseason3 wrote:Rolled my own from the directions linked from 4chan. Was easier than I thought. It's been about 24 hours and these are the best internals I've generated. A lot of trash along the way but these are pretty usable with a few tweaks.

These look very high quality, I can't seem to get anything even remotely near this on my own. Can you walk us through how you managed to get outputs like this either here or in Jim's SD discord? Did you get these through prompts alone or img2img? Did you use any inpainting to smooth out these results? What kind of tags did you use? What checkpoints/models did you use, did you use one of the original models in the voldy guide or did you also make use of the NovelAi model leak?


I used the stability diffusion 1.4 model in the guide to start.

Next I did do quite a bit of img2img work to get those. I started with the attached image from a well known photomanipers asset pack. the included Eskoz bg was helpful in getting a good start.

Using that as a base I just adjusted the settings to let the AI get creative and make new scenes from it. My base tags go something like this:

positive
masterpiece, best quality, masterpiece, high resolution, 4k, unreal engine, photo, photorealistic

negative
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, artist name

Those are what I always applied. From there were a lot of bizarre smatterings of tags to get weird results. I can't remember all of them but these were ones that I do remember yielding interesting results:

flesh cavern, mine, large cavern, meat walls, flesh walls, muscle texture, flooded mine, slime, slimy

There were more but I didn't save the prompts I used, sorry.
Attachments
scene00416.png
User avatar
ryanshowseason3
???
 
Posts: 2479
Joined: Thu Jan 05, 2006 12:00 am


Previous

Return to General Vore Discussion

Who is online

Users browsing this forum: DoomXBio