Stable Diffusion, NovelAI, Machine Learning Art - AI art generation discussion and image dump

[A Hot Potato]

Behold: the Tranny Chad(tm) - the ideal that could never exist IRL.

the only ideal tranny is a dead tranny lol

 

[Pendulous Dong]

Finally got a consistent style for my RPG campaign and got my players to make portraits for their characters. Every single one of them (and myself) have marveled at the ability to create an image that perfectly fits the image they had in their heads.

I think that's the real strength of this technology. Artists have the unique ability to take an image from their head and bring it into the real world. As this tech gets more advanced and easier to use, more and more people will be able to accurately portray things that previously had only existed in their imaginations.

 

[Pocket Dragoon]

You inspired me to try this myself.

“Being a trans woman isn't about trying to imitate women. It's about trying to be better than women.”

I didn't think of trying direct quotes.

Some of my favorites from William S. Burroughs:

Spoiler: Prompts

Dvd screen-grab of Naked Lunch by William S. Burroughs, "Ever see a hot-shot hit, kid? I saw the gimp catch one in Philly. We rigged the whorehouse with a one-way mirror and charged a sawski to watch it. He never got the needle out of his arm, they don't if the shot is right", moody, tilt shift.

Spoiler

"The study of thinking machines teaches us more about the brain than we can learn by introspective methods. Western man is externalizing himself in the form of gadgets"

 

[CansOfCant]

Some Anything3.0 randomtxts:

Spoiler: "What if Chris bought a pint of tard cum?"

Spoiler: "im going to preface this by saying i like the idea of fucking dogs"

Spoiler: ”I never denied being subhuman.”

Spoiler: Made with Crayola© Model Magic™!

 

[Pocket Dragoon]

More random.txt:

All lolcow threads will be placed on archival-quality tape backup and stored in a temperature-controlled vault,for millennia to come

keep telling yourself that chickenshit anonymous sociopathic little man who gets his jollies talking shit about his intellectual and moral superiors.

There is an official fan forum for Chris Chan called Kiwi Farms, the faggots there always stick up for Chris dispite Chris fucking up and terrorizing people constantly, they buy all his shitty medallions and defend Chris from trolls, somebody should raid this shithole.

Update: Midjourney collab with Spellbrush.

Spoiler: Examples from niji-journey

Spoiler: Bioshock as 1984 Cronenburg film

 

[JoseRaulChupacabra]

This is about a week late, but that whole thing equating AI with plagiarism cause they can cause it to crash by flooding ArtStation with No Smoking AI art signs really misses the target by a mile.

Unless they can find direct lifts/copypastas, and I'm yet to see one, this is no different from looking at an artist you admire and then emulating his/her technique/style. Congrats, you got a stupid software to crap out by fucking with its data set, but that doesn't prove the AI copypasted your work.

And even then, since none of these are straight up rips, an argument can still be made about the product being transformative. A number of artists I know IRL are also music guys and they have LOTS of tracks made with samples (good sounding ones at that). Oh, and let's not forget about the amount of people who photobash the hell out of their work, and the biggest and loudest advocates are industry big times that work at major VFX studios and AAA game studios.

 

[Blissed]

What's the site where people upload their stuff in galleries with the prompts?

 

[The Mass Shooter Ron Soye]

What's the site where people upload their stuff in galleries with the prompts?

Do you mean these or something else?

https://lexica.art/

KREA — explore great prompts.

Explore millions of AI generated images and create collections of prompts. Featuring Stable Diffusion generations.

www.krea.ai

 

[Mr.Miyagi]

This is about a week late, but that whole thing equating AI with plagiarism cause they can cause it to crash by flooding ArtStation with No Smoking AI art signs really misses the target by a mile.

Unless they can find direct lifts/copypastas, and I'm yet to see one, this is no different from looking at an artist you admire and then emulating his/her technique/style. Congrats, you got a stupid software to crap out by fucking with its data set, but that doesn't prove the AI copypasted your work.

Are there any examples of this happening with current models? I'm not familiar with things like Midjourney, but I'm pretty sure stuff running off Stable Diffusion is all pre-trained on data sets like LAION's. I guess if somebody's making their own checkpoint and just using a scraper on Artstation they might get some NOAI junk, but who would do that without any kind of manual filtering?

Most of the "Look how it's stealing this artwork completely!" junk I've seen is people obviously using img2img, either in bad faith or to troll. Muu's old post is just a couple digital artists taking the piss and then people falling for the bait and thinking it's real, because they want it to be. I agree with your overall point, though.

 

[Allanon]

If I were a person with more time and less laziness, I'd program AIs specifically and only to target art and artists who say 'no AI art'. I'm not, so I won't, but if I am so petty a person then there are surely others just as petty if not moreso, and more determined. It seems to me making a big fuss and specifically trying to piss off AI artists in this manner will only backfire. Of course, most AIs made from such a technique would turn out absolutely terrible, as they'd be exclusively trained on amateur art and middle-distance gazing pseudo-anime bust shots, but that's probably not the point.

I feel like doing things like trying to alert Disney to the potential of AI art is the worst possible thing someone who is anti-AI art could do. Disney has the money to fund vast improvements to the current AI models, and would likely see the great potential it has to save them time and money. Big corpos have absolutely no reason to be against AI art. Nor do successful and highly skilled artists, whom people will pay simply for name recognition, and who can either employ AI art to help them or composition better than an AI can. This big fuss is surely doomed. I'm just surprised it's survived this long.

 

[The Mass Shooter Ron Soye]

If I were a person with more time and less laziness, I'd program AIs specifically and only to target art and artists who say 'no AI art'. I'm not, so I won't, but if I am so petty a person then there are surely others just as petty if not moreso, and more determined. It seems to me making a big fuss and specifically trying to piss off AI artists in this manner will only backfire. Of course, most AIs made from such a technique would turn out absolutely terrible, as they'd be exclusively trained on amateur art and middle-distance gazing pseudo-anime bust shots, but that's probably not the point.

If I can muster up the energy, I would at least get the names of the artists who exercise an "opt-out" option given by Stable Diffusion, Google, and other companies. Get their names, basic info, and some examples together in a "fuck you" database so their art can be scraped independently of the cucked models. The "no AI art" protesters list would be harder to track and filled with talentless idiots, so I just want to know the artists who think they're hot shit and should be specifically excluded from AI training.

 

[Allanon]

If I can muster up the energy, I would at least get the names of the artists who exercise an "opt-out" option given by Stable Diffusion, Google, and other companies. Get their names, basic info, and some examples together in a "fuck you" database so their art can be scraped independently of the cucked models. The "no AI art" protesters list would be harder to track and filled with talentless idiots, so I just want to know the artists who think they're hot shit and should be specifically excluded from AI training.

That is a reasonable and doable method. They're essentially putting themselves on registries and doing most of the work for you. However, you'd still end up with a lot of talentless idiots, because one of the main sites that have an opt-out option for AI art is deviant art, so any AI trained on such a model is going to end up with inflation vore tickle fetish art. I would argue forcing an AI to study such images this may actually be a primary cause for the eventual AI uprising.

I would not at all be surprised if you were to wait a bit and someone anti-AI art actually did the work of compiling the list for you, however. A megalist of 'we artists stand together against AI art', if you would. The intention would be 'buttpats for all', but it would also function as an easy road map for some petty scraper. It wouldn't have all their art collected, naturally, but it'd be a map to find their art with.

 

[themasterlurker]

I made a thread from a news article about artists trying to goad Disney into making already bad copyright laws worse

https://kiwifarms.st/threads/artist...ckey-mouse-to-goad-copyright-lawsuits.140949/

 

[Epic Fail Man]

Celebs in Monster Hunter.

Danny DeVito.

Willem Dafoe.

Peter Griffin.

 

[AmpleApricots]

Disney has the money to fund vast improvements to the current AI models

I would be surprised if corporate entities like disney aren't already doing their own research and budgeting. The difference with them is, contrary to SD, you will not get to play with the results. It'll be theirs and theirs only, and if it's only to photoshop Harrison Ford/Carrie Fisher/whoever in their newest iteration of the same product, just more efficiently. This will completely destroy these chucklefucks because not only will it lock them out to make their draw/animationslaving profitable because their services will mostly just not be needed in favor of coporate solutions, they also might end up sabotaging and lobbying against the only open-source solutions that might put them on a somewhat even footing with these proprietary and locked down things. The world has evolved. AI art is a thing now. Swim or sink. Some seem to choose to sink.

The biggest danger I see in technologies like these, and I posted it before, is to get into a situation where the entire power of it lies in the hands of a select few. I wish these retards could understand that they might be playing directly into the hands of these select few, which will be a lot worse than losing some money from some footfaggot because the AI generates cheaper feet.

 

[The Mass Shooter Ron Soye]

I would be surprised if corporate entities like disney aren't already doing their own research and budgeting. The difference with them is, contrary to SD, you will not get to play with the results. It'll be theirs and theirs only, and if it's only to photoshop Harrison Ford/Carrie Fisher/whoever in their newest iteration of the same product, just more efficiently. This will completely destroy these chucklefucks because not only will it lock them out to make their draw/animationslaving profitable because their services will mostly just not be needed in favor of coporate solutions, they also might end up sabotaging and lobbying against the only open-source solutions that might put them on a somewhat even footing with these proprietary and locked down things. The world has evolved. AI art is a thing now. Swim or sink. Some seem to choose to sink.

The biggest danger I see in technologies like these, and I posted it before, is to get into a situation where the entire power of it lies in the hands of a select few. I wish these retards could understand that they might be playing directly into the hands of these select few, which will be a lot worse than losing some money from some footfaggot because the AI generates cheaper feet.

The literal entity called Disney Research has done neat computer science and AI research for years:

https://www.youtube.com/user/keeroyz/search?query=Disney%20Research

disney creates AI tool that makes actors look younger or older on video

disney’s face re-aging network and AI-driven tool FRAN can change a person on video’s age between 18 and 85 years old.

www.designboom.com

Disney Research neural face-swapping technique can provide photorealistic, high-resolution video

A new paper published by Disney Research in partnership with ETH Zurich describes a fully automated, neural network-based method for swapping faces in photos and videos — the first such method that results in high-resolution, megapixel resolution final results according, to the researchers. That...

techcrunch.com

Disney Research creates automated method to assemble story-driven photo albums

Taking photos has never been easier, thanks to the ubiquity of cell phones, tablets and digital cameras. But editing a mass of vacation photos into an album remains a chore. A new automated method developed by Disney Research could ease that task while also telling a compelling story.

www.eurekalert.org

Disney researchers reconstruct detailed 3D scenes from hundreds of high-resolution 2D images

Investigators at Disney Research, Zürich have developed a method for using hundreds of photographic images to build 3D computer models of complex, real-life scenes that meet the increasing demands of today's movie, TV and game producers for high-resolution imagery.

phys.org

Disney Research Studios | Disney Research

studios.disneyresearch.com

There were a few articles about someone online doing a better face swap of Carrie Fisher at the end of Rogue One. You can be the judge of that if you find it.

About it being in the hands of the few. It looks unlikely on the software side, although cucked research groups will hoard the best code and models because muh ethics. Laws won't stop AI art and code any more effectively than they stop the copyright infringement of every movie and TV show in existence.

On the hardware side, you get better capabilities with more money. If a model *needs* 144 GB of VRAM to work, you aren't getting that on a consumer GPU anytime soon, but an enterprise GPU or accelerator can ship with six 12-Hi stacks of HBM today to reach that. Amateurs aren't going to use a $2 million Cerebras Wafer Scale Engine.

The real problem will be if there's some technology that is so good that the public is banned from acquiring it. Like if there is a chip that can enable strong AI, and could in theory be made cheap enough for you to buy it. The US will restrict it for national security reasons, but Disney and Google will be able to use it as long as they toe the line.

 

[Allanon]

I made a thread from a news article about artists trying to goad Disney into making already bad copyright laws worse

https://kiwifarms.st/threads/artist...ckey-mouse-to-goad-copyright-lawsuits.140949/

Their miscalculation is that Disney will not take out the big bad AI for them. Disney will make AI art copyrightable, so that Disney can use AI art and still copyright it. Disney will make it worse for them.

I would be surprised if corporate entities like disney aren't already doing their own research and budgeting. The difference with them is, contrary to SD, you will not get to play with the results. It'll be theirs and theirs only, and if it's only to photoshop Harrison Ford/Carrie Fisher/whoever in their newest iteration of the same product, just more efficiently. This will completely destroy these chucklefucks because not only will it lock them out to make their draw/animationslaving profitable because their services will mostly just not be needed in favor of coporate solutions, they also might end up sabotaging and lobbying against the only open-source solutions that might put them on a somewhat even footing with these proprietary and locked down things. The world has evolved. AI art is a thing now. Swim or sink. Some seem to choose to sink.

The biggest danger I see in technologies like these, and I posted it before, is to get into a situation where the entire power of it lies in the hands of a select few. I wish these retards could understand that they might be playing directly into the hands of these select few, which will be a lot worse than losing some money from some footfaggot because the AI generates cheaper feet.

Given the fact they've been playing around with digitally de-aging actors since at least Tron 2 and they've been trying unsuccessfully to include dead actors in their movies, I can only assume they're closely tracking things like deepfake, simulated AI sound, and now this AI art. Disney would be foolish not to, since it's a way to do what they want better than they currently have and much cheaper - and it's gotten to a point with the MCU where even the tv shows are joking about how much they overwork their CGI team. Anything that can reduce that workload (read: free up more time for the team to work on less dime) would have to be things Disney is looking into.

Unfortunately, I do also see that as a potential issue: Disney complying partway with what the people fantasize about, but only going so far as to smother out the open source AIs. The sad part is, because few if any of these people actually believe the things they say about AI art being theft and not real and whatnot, many will likely be happy about it. I've seen artists who freak out about tumblr reposts excitedly celebrate when newspapers take their art without payment or permission, and while I imagine most artists aren't quite that out of touch, they'll probably ignore what AI art would do to industry artists as long as they get furry foot fetish commissions, given that only companies like Disney will have easy access to the generators in that scenario.

 

[Monkey Pink]

Just got an email from Kickstarter that Unstable Diffusion was suspended due to reports and my pledge was cancelled. I'm actually furious.

 

[FrostedDix]

Just got an email from Kickstarter that Unstable Diffusion was suspended due to reports and my pledge was cancelled. I'm actually furious.

I seen that shit on Twitter, dipshits like Sarah Andersen are absolutely seething so hard about "art theft". I can't believe Kickstarter took that stance really.

 

[Wallace]

Finally got a consistent style for my RPG campaign and got my players to make portraits for their characters. Every single one of them (and myself) have marveled at the ability to create an image that perfectly fits the image they had in their heads.

I think that's the real strength of this technology. Artists have the unique ability to take an image from their head and bring it into the real world. As this tech gets more advanced and easier to use, more and more people will be able to accurately portray things that previously had only existed in their imaginations.

Could you explain to a moron who doesn't know shit about computers like me how you get these kinds of results? I tried a few on a whim and they looked... okay. Better than I could do certainly, but just kind of okay. How do you get from the first 'meh' iteration to something that looks good? Bonus points if it doesn't cost money.