How did you get past the filter

>>4773 i used this gui of the open source code that had the filter turned off, also made things easier https://grisk.itch.io/stable-diffusion-gui

>>4772 interesting, have more? I don't have a graphics card

>>4775 unfortunately no, it's really difficult to get anything else similar to this from the ai. i honestly only got really lucky getting this good of a pic because any other attempts didn't include a dog or porn at all. i was able to get this one anthro gsd though, don't know what's up with the weird clothes on him if anyone else would be able to try it out and get better results that would be great!!

>>4777 oh god..

What prompts are you using? "Husky blowjob" is the only thing I've tried that seems to give anything close. Other stuff just seems to ignore part of the prompt, but even still, the most I can get out of it are deformed messes or disembodied dicks.

>>4782 the prompt that i used for the first pic was 'gay porn bestiality german shepherd dog'. i went for a really simple prompt and iirc it was the 3rd generation attempt. all the others after never came close and were similar to yours, unfortunately. i also tried using the same seed to try to generate variations but that only made it worse and admittedly horrifying i also found that trying to be more specific in your prompts tends to make the ai deviate away from generating actual porn, at least when it comes to involving animals

>>4783 That prompt style does seem to give more explicit results, and like you, it spit out a good one pretty early and then produced irrelevant or poor quality ones from then on, probably just a coincidence.

>>4784 this is a really good one. i've still been trying to generate stuff but no luck so far

Can someone make one of an unrealistically large object getting shoved up a dog's ass? Like maybe a man shoving his head up a husky or german shepherd's ass, or maybe the dog is pooping out a football or watermelon.

Can't get anything good myself, blowjob is the only tag that has made anything even remotely zoo

You must use your words...thesaurus style and such. Some pretty crazy things come out of stable diffusion...

>>4864 >>4865 Could you give us some examples of the prompts you use? I've had some come close to yours, but not quite as good.

I have used so many, I couldn't say what kicked out what... I just substitute 'dog' with 'canine' or a particular breed... 'fuck' with 'sex with', 'dominated sexually', 'intercourse', 'coplulate', etc. As well as switching the order of everything around or saying it slightly different, i.e. 'man fucks dog' being 'dog fucked by man'. I found best results without actually having 'man' or 'woman' in the prompt...just 'fellatio dog' or 'sex with dog'...gotta get creative.

>>4868 I have done thousands at this point, dalmation poodle and chocolate lab seem to give the best results as far as breeds go, although you will get bbc with chocolate lab

Has anyone tried something more simple, such as dog pussies or mare udders?

Of course...in both slang and proper terminology. One can train the whole setup on a particular topic, but I don't happen to be well versed in linux enough to do figure out how to do so yet.

>>4915 The closest I got was dogs with human pussies very rarely

where's the dolphin porn?

>>4923 hmnhh..I guess this AI needs some training

I found an interesting use for the Stable Diffusion AI. You can use it to turn dogs into wolves. And maybe other species too. What I did was chose a zoo photo with a dog that's already kinda wolflike, then made a square crop the dog's head, then uploaded that to theartbutton(dot)ai, then used a prompt like "arctic wolf head" or "timber wolf", then played around with Image Strength (between 40%-70%) and different seeds until I got a good-looking wolf head in the exact same lighting and pose as the original dog, then photoshopped the newly generated wolf head back over the dog's head. AI wolf vs original dog for comparison. I like the 1st one the best. If you make your own, post em here.

The main issue with AI models is that they need to be trained on the type of content you want to generate. And I doubt that zoo porn is part of the training data set.

>>4999 I saw on reddit someone used the open source stable diffusion to create a bunch of pretty impressive furry porn because they removed the folder that had all the data stored for image training and trained their own version of the AI exclusively on the entirety of e621. Maybe we can do something like that but with beast stuff?

>>5000 Yes it is of course possible, if you know how to train a neural network and you have a good data set. And by data set, I mean more than just a bunch of pictures. They need to be tagged, filtered, etc. That alone is a lot of work.

Not to mention the hundreds, if not thousands of hours it would take unless one had a few grand into their computer, and another 5-10k into GPUs.

perfect just what i needed even though i dont fap to this shit is there some kind of spambot for twatter/plebbit im going to use it to flood the PMs of self righteous wannabe anti-zoo pro-lgbt woke accounts to tech them a lesson for messing with gamers

>>5081 I just choked on my beer laughing at your comment... On one hand, the irony of you reading a forum dedicated to zoo porn and saying that you don't fap to it, but then to say you're going to use it as a weapon to spam peoples message boxes... I love you. I don't know you but I'd give you a blowjob as thanks just for these comments.

>>5081 Wow, we've got a baddass over here.

stuff has been improved since i started this thread so here's some stuff i managed to generate recently. i've only been able to get blowjobs though. it's still not perfect but definitely an improvement

more pics

>>5689 Woah! it's starting to be better! let's gve more 2 years to see how it's will be this AI

>>5690 I need more of the horses! Also is it able to do two dogs/ horses at the same time?

>>5690 What model did you use for these?

>>5719 i'll try to generate more when i have the time. also haven't tried doing more than 1 dog or horse, it was hard enough getting these results out of the many deformed ones tbh

>>5770 homoerotic by gaydiffusion. it's on civitai

Far from good, but the best I can get so far. Last one kinda looks like a man knotted to a german shep going butt to butt if you use your imagination. I think to get what we're looking for someone is gonna have to download a shit ton of zoo porn, make a massive list of tags, and creat their own model.

Not a sick freak to be clear but can this thing be used to generate bloody zoosadist content? Is the quality good enough to alarm law enforcement on my "accidentally leaked video" smear campaign on a hacked account? As twisted as my evil plan sounds im gonna plot some serious revenge on my old bullies by utilizing face swapping algorithms and weponizing plebbit's self righteous vigilante mob mentality for my own nefarious agenda. Now imagine the implications of suddenly seeing your target's face on a viral liveleak video and him being beaten up brutally by strangers for no reason at all in the confort of your own home. Man i could make a lucrative but highly illegal real life hitman service out of this for monero. What do you think? Perhaps i should name it KAAS karma as a service while appearing as an independent justice centered NGO on the surface

>>5837 go kill yourself. you're mentally unstable.

>>5837 >but can this thing be used to generate bloody zoosadist content? Probably not because it can only work off images that are already in its dataset, and there's not gonna be zoosadism in public datasets like LAION-5B. If you wanted to make your own dataset that has it, you'd need a lot of that kind of content to train it with. Though I guess probably butchery images or surgery images are in there, so that might give similar results (a prompt like "man holding up a dog while it's having open heart surgery" to make it look like someone cut open a live dog). >plot some serious revenge on my old bullies Best revenge is stop letting them live rent-free in your brain and just move on; delete all memory of them from your brain >accidentally leaked video Video generation is a bit off yet. Takes lots of processor power just to generate still images as it is. >by utilizing face swapping algorithms and weponizing Everyone already knows image generation/deepfaking exists, and such content is easily discernable if not by eye, then by counter-ai 'ai-generated output detectors' that are now being developed (for example, "openai-detector" can tell if a sample of text has been generated using chatgpt). Not that ai even adds anything new to the piture: faking has always been possible manually using video editing, impersonations, visual effect, splicing together different videos, just plain lying, etc. >>5689 These look good!

>>5870 Zoosadism is a federal crime to possess or share. Can we please change the subject before some dumbass posts some and I have to ban him. I don't wanna ban people guys. Be cool cmon.

>>5871 Dude, relax. No one here's into zs, that dude seems to just want revenge, any revenge. But I hope he takes my advice to just say fuck'em and move on. Making up worst-case scenarios in your own head and worrying about them happening is a self-destructive habit that makes you see peril around every corner. Better to just relax. >>5690 Can you try to make some of these with a wolf? Arctic wolf or timber wolf maybe.

>>5871 no it is not. crush content is.

>>5874 Yeah good point bro. Sorry. >>5876 Yes and no. Yes it is only "animal crush" thats illegal, but the definition of animal crushing is the issue. Its broad to the point of including any animal abuse. (1)the term “animal crushing” means actual conduct in which one or more living non-human mammals, birds, reptiles, or amphibians is purposely crushed, burned, drowned, suffocated, impaled, or otherwise subjected to serious bodily injury (as defined in section 1365 and including conduct that, if committed against a person and in the special maritime and territorial jurisdiction of the United States, would violate section 2241 or 2242); (2)the term “animal crush video” means any photograph, motion-picture film, video or digital recording, or electronic image that— (A)depicts animal crushing; and (B)is obscene So yes and no.

Legality is irrelevant. That shit has no business being here.

>>5878 Yeah thats a better point.

>>5877 No it is not. Nowhere in there is animal abuse mentioned. It's talking about crushing. "Serious bodily injury" does not include sexual acts, which are covered under a separate law. You have no reading comprehension of legalese. Come back when you grow a fucking brain.

>>5885 Incorrect.

>>5877 You must live in Nazi England

>>5907 Thats US federal law anon. They dont care about that in ol blighty. https://www.law.cornell.edu/uscode/text/18/48#:~:text=It%20shall%20be%20unlawful%20for,jurisdiction%20of%20the%20United%20States.

>>5877 >including any animal abuse And that's a good thing! Zoosadists should all die in a fire.

>>6374 yeah pretty much

I mean to be clear everyone involved in fucking an animal is going to hell, but also yes fuck animal abuse thats fucked up.

>>6376 Based

ControlNet is pretty great

Stable diffusion is capable of generating surprisingly good zoo art. Best with only one subject though, trying to throw a dick into the mix results in some weird shit.

>>6771 what models are you using to get high quality cookie like that?

>>6772 fluffyrock, probably

>>6772 https://civitai.com/models/17649/bb95-furrymix with the following prompt: e621, feral (dog), female, solo, from behind, canine vagina, canine pussy, hindquarters, anatomically correct, photorealistic, good quality, best quality, realistic fur, tail NEGATIVE: (solo:1.4), (worst quality:1.4), (low quality:1.4), (cropped head:1.4), (blurry), blue nipples, bad-artist, boring_e621, bad_hands, bad_anatomy, worst quality, low quality CFG somewhere between 7 and 12. Works best with image size set to 720x720 or so

>>6778 Thanks for giving me a starting point! That model is way better than anything I've tried before. e621, feral (wolf), female, solo, laying in flower meadow, laying on back, canine pussy, looking at camera, spread eagle, anatomically correct, photorealistic, good quality, best quality, realistic fur, tail same negative prompts

upscaling makes the faces really pretty but ruins the vulva, sigh

sometimes it cranks out some really good ones

>>6905 i love how you can tell its being trained on truegrave art because of the logo in the bottom right lol

>>5798 Jesus this almost took me 3 full minutes to realize its fake now imagine what someone like mossad could do with this kind of weapon provided theres enough funds. Time to bribe the OpenAI devs for my newest false flag shooting gayops modeled after google street view Now on a serious note anybody got a model trained off zoox18/luxureTV videos? for >>5000 these prompts are gonna be fun as hell to generate >>6778 Not a pedo or anything, BUT This is your warning

what the? what did i do?

>>5870 >revenge Nah that was a long time ago its just an example scenario now lets say if i were to handsomely get back at one of my employees for leaking dirty company info all i have to do left is put up fake wanted pedo flyers right after orchestrating the deepfake and the rest is history like the /i/ guide intended nobody is gonna think twice before beating him up its nearly foolproof. I also heard that FEDS brand important targets as sex offenders without any proof effectively locking them out of society is this true? They tend to go pretty extreme lengths as another altchan mentioned. >processing power Remember that time when PETA fags made a fake 3D video of a cat getting beaten and the jewtube outrage mob thought it was real before they announced it later on? AI wasnt even a dream back then secondly most of the shit we see on liveleak is 480p tier filmed on a crappy dumbphone so all i have to do left is to add a downscale filter to foil their countering AI as it was designed to only work for high quality images Or just simulate the room traditionally on SFM/unreal engine then we can put the faces on the characters while letting AI do the finishing touches plus you can also customize the movie scene as needed for it to be truly convincing that's the plan after all we pretend to be fake NGOs/journalists smearing shit under the guise of corrupt exposure. That'll teach those twitter vegans who's boss.

>>5798 Jesus this almost took me 3 full minutes to realize its fake now imagine what someone like mossad could do with this kind of weapon provided theres enough funds. Time to bribe the OpenAI devs for my newest false flag shooting gayops modeled after google street view Now on a serious note anybody got a model trained off zoox18/luxureTV videos? for >>5000 these prompts are gonna be fun as hell to generate >>5871 Hmm if i were the owner illegal snuff producer perhaps i could use this weird new tech to make it appear is if all my films were fake incase cops knocking down for plausible deniability but im not that evil IRL Also fun fact theres a news article about women on ordinary onlyfans using AI to hide their real faces allowing them freely whore away without fear of getting doxxed and on the other hand one American student decided to kick it up a notch and created full fake girls and earned up to 40K dollars worth of subscriptions thanks to the deepfake at first he never even expected those to even take off at all but what do you know? it just works in the end i guess this is the future of the porn industry now not demented in the head or anything just my hypothesis on how to properly weaponize this for your own agenda i should probably stop giving feds more ideas here i just stumbled upon this thread out of nowhere i dont really use this site at all sorry about that earlier

>>7032 >>7038 You're borderline asking for CP content, bro. Ask on a different board.

who has a free ai generator that doesnt need a fucking phone number to use it

zoo porn

Figured out how to make some decent stuff, any requests?

>>7635 I wanna see a dog beating some chick in bdsm leathers. Like maybe the dog has a whip even.

>>7636 XD, well I can tell you that would be extremely difficult to make. Was thinking more like what I posted I should have clarified, I can change the animals pretty easily.

>>7637 oh then uh. Idk. Never seen a giraffe blowing a dude before. Maybe a giraffe and a couple of uh... Otters. yeah why not. Can we get a seal in there too?

>>7638 I'll return when done :)

>>7639 Well all fucking right. Thanks anon.

>>7635 Would it be possible to make a similar one to this, where the horse has a huge facial and is drooling onto the dick? Alternatively, if you could make a dog with a long tongue drooling all over a dude's asshole, that would be appreciated as well.

>>7642 This is the craziest shit ever. I regret nothing. Thank you anon.

>>7645 Working on the giraffes rn, trying to get their tongues to look more giraffe like is a bit hard

>>7646 Ive never seen a giraffe tongue. Prolly pointed like the horses in your first pic tho id guess... Shit now I gotta go find out.

Well I found this.

>>7647 They are long and purple.

>>7906 are you sure you're on the right board, anon?

>>7908 Emilia is a bitch, so I'd say he is in the right place.

>>7635 Could you do a male dog in bondage?

>>7942 Uh. What.

Has anyone here tried building some tagged/described dataset to train a LoRa (or an SD checkpoint)?

>>7976 have not tried yet.. rebuilding collection now, and planning to tag a dataset soon. any recommended reading / examples?

My dog fuck a bear

I am the meeting and I want to see animals f** each other

cuh

>>7642 Not up to this quality, but I managed to get it to gen this.

>>8369 I would like to find a model that can generate real images zoo porn

I've been fine-tuning the SDXL 1.0 model and am starting to get some good results maybe.

Here some of my personal creation, still learning about SD / Ai, so expect not that much of amazing generation picture Starting with Rabbit

Owl

Sheep

Beetle -1/2

Beetle - 2/2

Cockroach

Mantis

Mouse

Lady bug

Mantis - 2/2

Snake

Chicken

Turtle

Bear

Goat

Spider

Donkey

Frog - 1/4

Frog - 3/4

Frog - 4/4

Cat

Last chicken

That's all for now. Still learning about it and I really want some guide for more, like, put more than 1 animal per picture. I tried but the result are all abomination lol Anyway, I hope you like it

>>8646 Incredible stuff anon, thank you for putting so much effort into learning this and sharing the fruits of your labor. Keep doing what you're doing, and maybe render some more rats and mice if you're feeling generous lol

>>8651 Here we go, more mice for everyone

>>8644 >>8624 >>8623 These are amazing! I hope to see more of these animals in the future.

>>8653 Could you please do a fox, and with anal penetration if possible?

>>8660 Here we go

>>8664 >>8663 fuck yes, thank you, that is so hot!!

>>8663 looks cool

>>8657 Thank you anon! I always love seeing animals dripping cum like this mouse does

>>8623 >>8630 >>8644 Really amazing stuff! Did you train the model? Is it a merge of Loras of Zootopia-style animals and porn images? Is there a checkpoint you can share? So many questions.. lol

>>8676 No trained Lora/Model was used, I only used a Model checkpoint with prompt and that was what we have heeh The Model was Furtastic, you can find in Civitai

Some bing stuff. By the way I love facials.

Imagine what happened in that picture. Multiple guys having fun with these dogs. And finally blowing her huge Loads of cum on there.

>>9110 I fucking love this one! Keep it up!

>>7635 Why not aquatic stuff like Dewgong, why not try cows and their udders? I do remember one anon that used two humped camels on his generations of lewd so far. Sorry for the avalanche but... today I'm inspired as fuck

>>9055 Always drooled at the concept of a bitch getting gangbanged, but the sight of two getting fucked side by side is diamonds

>>9116 Yeah. I want to see a bukkake party, where 10 guys cum in a row on a cute doggo.

This is so fkn Gay, this should be a gay thread rather than zoo.

>>9181 Ooh, what did you use for the second one? Something like "sat on melted ice cream"? By the way, if you're looking for other white slimey liquids, "watery oobleck" also works.

>>9201 I used "lifting tail" and "yogurt dripping off of nose" and it got super lucky and put some under the tail too

>>9203 Fuck yea, that's awesome.

horse pussy

for the less technologically minded, or those too lazy to use a local ai (like me): the latest bing dall-e ai can.. ALMOST produce some good stuff. extremely hard to get highly visible genitalia but its still sexy in a vague way. best part is that its free and very easy to use. https://www.bing.com/images/create you use tokens to generate images otherwise it might take a while. however theres a silly bug to work around it, when you run out, click the top right settings, go to search history, hit clear history (might need to validate account), wait about 10 seconds then go back to the link. should reset tokens back to 15. also try to avoid warning messages, too many = a temp ban. you need to change the prompt if you get this message. if you get the default dog image or something, it just means it detected an innapropriate image but theres no real consequence for it, just keep prompting until something goes through. or play around with the prompt and it might become more lenient.

another quick thing: misspelling words sometimes works. i found "genitalia" (correctly spelled) gives a warning, but "gentialia" went through, and i believe the ai basically knows you mean genitalia and will attempt to provide it.

>>9225 Cool, thanks for the token tip and the mispelling tip, it beats using weird synonyms. If you trigger the block (after about 15(?) blocked term attempts in an hour), its just a temporary 1 hour ban from the service. No biggie. Also, if you wanna get rid that perfectly-lit 'ai look' you can get creative with how you start the prompt, for example "wildlife documentary screenshot of...", "movie screenshot of...", "comedy movie screenshot of...", "80s sitcom screenshot of...", "fish eye lens photo of..."

I want to see Putin gettin fucked in the ass - anyone can do that here <3 ? Woiuld be so hot.

>>9482 Noice

dog having sex

>>9559 does anyone know the models / prompts for these? in my experience this is beyond what I ’ve seen/gotten current genAI, this is very interesting

>>9601 https://booru.plus/+sdg

>>9601 answering my own question after finally finding an original that still has metadata: [(by thebigslick), (by dimwitdog:0.8), (by twistedscarlett60:0.8), (by foxovh):(by slugbox:0.8), (by funkybun:0.8), (by whisperfoot:0.6):0.6], [(detailed_background:1.4), (high_contrast:1.2), (vibrant_colors:0.8):0.2], [(intricate_detail, realistic, real, photorealism), <lora:add_detail:1>:0.2], feral, doberman, looking at viewer, ((bedroom eyes, sexy smirk)), pawpads, anatomically correct, canine pussy:1.1, anus, outdoors, bent over a chair with feet up, dripping pussy juice, shiny genitals, glistening glizzy, rear_view, Negative prompt: bwu, dfc, ubbp, updn, mangled_hands, blurry_eyes, (extra tails:1.2), (extra ears:1.2), (extra anuses), (extra horns), (extra fingers:1.2), (extra horns:1.2), (cutie mark), glans, faded_colors, sepia, greyscale, monochrome, split hooves Steps: 40, Sampler: DPM++ 2M SDE Heun Karras, CFG scale: 10, Seed: 2326344199, Size: 768x640, Model hash: 5be73da75b, Model: QuEasyFluff, VAE hash: c6a580b13a, VAE: vae-ft-mse-840000-ema-pruned.ckpt, Denoising strength: 0.4, CFG Rescale: 0.25, Hires upscale: 2, Hires steps: 14, Hires upscaler: 4x-UltraSharp, Lora hashes: "add_detail: 7c6bad76eb54", Version: v1.6.0

>>9559 >>9616 >>9627 These are incredible! how do you make them?

What are people using these days? I heard Stable Diffusion doesn't work anymore for NSFW

>>9814 >posts in recently active thread full of NSFW gens using StableDiffusion asking if NSFW no longer works so do you just keep morons for company, or did you brew this one up yourself?

>>9820 I guess I keep morons for company, because that's what I read from someone somewhere. Here's a bing one I guess, to contribute

can we get more cats?!

>>9808 >>9809 very nice, what's the prompt?

I made a shit ton of cat ones

and that's it

>>9923 Do some gay cat anal

Dog fuck girl

>>9940 I don't do gay ones.

Can anyone make any zoo ai pics that are based off of some actual porn studio vids Casting Couch, Mike Adraino, Swallowed, Throated, Blacked, Step Bro fucking step dog,

Why not try dolphins or cows? why not even camels or alpacas? I think we have seen enough canine and equine on AI zoo

>>10124 yes, but only if they train the model them selves. They would need lots of source material to teach the AI.

>>10142 Holy shit that's AI? What model are you using!?!?

>>10142 Damn! It's so great. I bet it took a lot of time to get result like this. Could you make similar with female shep?

Are there any AI animations? I would love to see some :)

EXTREME

Bing patched dick generation. For now you have to make typos in it if you need animal cock or butt and not only "not" prefixes.

fuckn civitai no longer works. Or maybe just not for free

It is neither expensive nor difficult to get things running locally on your own computer...I'm only running 8GB VRAM, and I've been training LoRa's, model checkpoints, embeddings...and then the image generation is the easy part.

>>10622 On the training part: Is it true that it takes on average about a week to start getting it to a style you want? Like if you fed the same style or type of art to train.

It's not really the actual training part that is time/attention consuming...that's not to say that the machine learning portion is quick - I'm running 24 gigs total shared VRAM, with only 8 gigs dedicated, and training time for a LoRa runs me usually under 10 hours. Depends on how many images and iterations it's running. Training a full model checkpoint I haven't done yet, because one can very likely come up with satisfactory results from currently available models and adding a custom LoRa or 2. What is truly a pain in the ass is getting all of your input images described and tagged fully and correctly. Textual inversion can get the process started, but it's not real accurate with zoo content.