AI is going to be our biggest game changer
This discussion has been closed.
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
The tech will certainly not be uninvented. But it could be regulated and large-scale centralized commercial exploitation could even get outlawed, for instance.
Perhaps other countries will become more intrusive in wanting to know your internationally acting corporation's innerts, and might start to raise taxes for ai being used instead of workers, based on randomly made up calculations, for instance, potentially similar for automation in general. But not only for the US, e.g. the EU has law in the making, which states "high risk applications" in the context of "ai" (probably not art, but things get regulated).
Depending on how it proceeds, such could lead to separated internets, economies, even more blocks, even if it's just a form of "economic war", after all... easily. Even for the US this is at least a question of timing, even if you don't care who or what can die with this or similar "technology". For this context it's important to understand, that ai only others can train, may give you an advantage, but in contrast to cloud software, of which the function may be well understood, and that you could exchange for other software at some cost, ai poses the double danger, both like what the cloud already means for reliabiliy, lock-in, and control, but also that you can't be sure, if you get the "export version" of an ai, rendering you a couple of % worse than for instance U.S. customers. Then you never know, if it stops being relevant the next day, and it's an even worse lock-in than updates. A working software for specific tasks can still work on-premise, an ai that is obsolete is useless from a certain point on. This is a similar potential, that might lead to regulation in some places, e.g. that you only can employ such systems, that can be trained on-premise, as well are constructed in a modular fashion (concerning specific parameters of it's use, not diving too deep here). Why would it raise to any power at all... likely due to "it works" for efficiency or enhancing, or replacing workforce (yes that's a huge driver in cost-management, anything else is naive to assume, look at tech companies laying off workers), at first including "hype", short term oriented commerce, meaning you can't keep up in a timely fashion, and you're out.
That's not only art, of course. I just want to illustrate ;p, where "this" could go, and that regulation is quite immanent in a few places.
You still need to understand, that training data has to be there at all. Companies will not be court ordered to have clean training data out of nowhere, they will try to have it clean-ish for their own interest. They'll filter too similar or generated-looking stuff out simply. But they'll then notice, that there is fewer and fewer applicable input, and that is the ruinous part. So walking into the big cloud service scenario blindly, could be called taking a huge risk for society. In the end you have created that "no illustrator" benefit, but your society stalls in the state of 2022 "forever". Of course that's the simplified cleanroom dystopia version of it. Not solving the question... hazardeurs?
Never need to .. if they don't want to... that's right. Though for a part, you could have that with a lot of cc licensed content already, e.g. training on it, or a little slower, digging it up manually each time.
(DISCLAIMER: Careful with any "classic" license like CC, do check for explicit consent, and check the official creative commons site for their take on training for generative ai. Things evolve, so might stances. I threw in the "could train on..." for fancyness, but i don't want people to take any risks based on random internet posts in general. Similar to copyright in general, you could in theory violate any attibution-based license with a generated image, so do check their stance and consent on it.)
(On a side-note: "The MP3 format ( legal or otherwise) outlived Napster" - mp3 has never been in question, and would have worked without napster. While mp3 may have been the best thing around, back then, the whole thing can not be compared to the current situation at all. MP3 just enabled sharing in the broad, due to slow internet connections, but it does not produce music on a button-press :). The internet allows fetching what fits into your connection within some time frame, perhaps, but that hasn't given rise to rethinking copyright in a fundamental way, really.)
This is a particularly good video from Sam Does Art explaining the issue of AI generation from the Artist's POV. An artist who created the art from scratch, literally from the blank pag, e has a different appreciation for the hard work involved in creating something, and why copyright protection really does matter ! No one is saying AI doesn't have a future but the AI companies need to recognize and respect the copyrights of the artist they scraped from.
As long as we're sharing info, this is an interesting article about a lawsuit against stable diffusion. It also efficiantly covers how the AI makes an image and the role of text prompts.
https://stablediffusionlitigation.com/
The only AI that I'm looking forward to is 3D mocap from 2D video! Imagine just how advantageous that would be for our hobby!
Have you seen this:
Hobby, ... profession... who knows. With knowledge about body physique, which already is present... certainly, eventually.
Carefully and/or skillfully crafted stuff will likely stay ahead for a while, but they might also need flexible tools to stay ahead with a then still reasonable amount of overhead as well. I'm confident, that such tools will exist. The medium-sized question remains, if it will shove money into one direction (<= number of big tech) , or if there will be actual competition (, which i rate to be more likely, at least for animation).
https://plask.ai/
I saw this video being referenced in the other video, whose maker/frontman tried to stay neutral.
From my side (some physics / information science / spoilers background), watching this one now, i must say, it appears to be very spot on, not leaning out of any windows. Subscribed, intellectually, algorithmically.
Speaking of windows, i feel the "it's there, accept it" part, especially with disregarding individual rights and sentiments, reminds me of the "official" Russian "handling" of the Ukraine "question", including current events. As if we had no choice. (No guarantees for current concepts from my side, though...)
well nobody has died or had their house bombed because I am experimenting with and learning how to use ai art generators
I also have not deprived anyone of their income
people should look at how far they can fall from preaching on their high horses because nothing happens in a vacuum
don't copy anything, don't use paid Russian Software like Wrap3D etc if you feel so strongly
I can not like what is happening in the world and say so but can still learn using free opensource software including stuff made by Russians and sleep perfectly well
I have openly said LAION 5B should not have been the training source and agree Stable Diffusion using that Checkpoint should not be used commercially
but I also will use it noncommercially in an educational way until they retrain an ethical database, I might die tomorrow, life is too short
Rest assured, you're not "the problem", neither is so the existence of Russians, whatever professions. I can't guarantee everyone to be out of trouble for publishing generated images in general. On the contrary, i would encourage people to have a look how it works and what it does, if you can. (Using a paid service already is another question, up to you...)
The LAION 5B question... i am not fully sure. I could envision actual publicly available science, both in paper and software, i.e. everything involved, but it gets sketchy with "non profit" being somehow transferred to "clearly commercial". I hadn't heard of those aspect earlier on.
"nothing happens in a vacuum" - of course, but depending on legislation and further, some people might be thrown into one (...), which should be the reason for some/most the critique on what's happening. I am not sure which part of lawsuit or critique to which project is fully justified or helping the case of societal progress, but i am 100% sure, that it's a terrible idea to argue, that everything should be just so available for any (large?) commercial endeavour.
"but I also will use it noncommercially in an educational way until they retrain an ethical database, I might die tomorrow, life is too short" - non-profit strictly (transitive-recursive, no tricks) would be a great opportunity, societally. It wouldn't take long to make scrapers respect a standard of "explicit consent only" as well as "no keeping removed images", so i don't see a societal issue with the technology itself (except for ignoring fundamental questions with destroying things needed for the tech to even exist, regardless of commercial use or whatever), but it looks like some people running such tech on the commercial side, have a different idea of how to proceed from here.
it is just I have been called a thief on facebook for sharing ai art by DAZ PAs who seem quite happy to model clothing resembling stuff from popular games and shows
I also know many who are also happily using ai too
and no I only subscribed to Midjourney one month last year, won't again unless things change
I am by no means enthusatically promoting ai art, I am firmly on the fence and balancing as best as I can
I won't embrace it or dismiss it I will dabble in it
there is way too much of a divide on this subject
[Hypnotic distorted smile / effects] "Are you sure Facebook is good for you?"
Obviously you will have many people creating popular stuff - even artists here do so, though in less blunt (violating) or say, in more smart ways than just 1:1 copying?
I could imagine copyright being handled differently in future, for less/non-commercial memes/fan-fiction. While that's another topice in a way, it's also tightly linked to the image-generation technology as presented now, because that clearly has that encyclopedic side to it.
this is why I am wary about that class action being mounted
I think it will have much futher reaching effects if it is not thrown out of court
and it is not the starving artists who will benefit, on the contrary they will get penalised for copying too
In theory, by my humble extrapolations, destruction-wise, many directions are outright possible to happen at this stage.
E.g. "strenghtening copyright on images" could lead to "similar" becoming the next entrance card to courts or at least automatic filtering.
But the result needn't be a catastrophe, it's not over yet. Not sure US- (or any?) courts are capable of salominic decisions, though.
No, I have not, and thanks for the link!
The tech is indeed here, yet I'm waiting for the tech to advance BEYOND greedy rotten rentware, and 3rd party websites that monitor your uploads, 'til I can buy/download a program, (and not some phone app) so I can load a youtube video or it's like into it and encode/convert to BVH for easy transfer to DAZ, we just aren't there yet, as there are too much bollocks to deal with via 3rd party greedsters/websites, that charge hundreds annually!
As even "FREE" means worthless shareware, when 80% of the functionality has been truncated unless you rent their over-priced "solutions," no, we still have to wait a bit before we can use the tech unfettered!
Yes, I forgot that I brought this up in this very thread, many pages ago, and thanks again for the link, I can't use the service but the link is still valuable for me as it illustrates just how close we are to being able to buy and download a single program/and/or have this tech integrated into Daz, rather than the **specific case usage that these companies provide as we're still in the commercial service phase of the tech and not yet into the private software arena! **(IE: Game development/industry animation)
I give it another year or so before the tech is common enough for use by simply downloading a Youtube video, loading it in the program, and converting it for use with Daz characters, and not needing a webcam or phone to upload to some greedy/and not so greedy websites, as even "free" solutions have the results severely truncated unless you rent out the tech!
Plus these sites/companies rely on ***G-rated SFW vanilla use-case scenarios as private use is not controlled so easily!
Thanks again for the link, Wendy!
***(And no, I don't mean x-rated adult fare, even PG is good enough.)
I mean, IMAGINE a DS plugin that will allow for loading a video, and converting it for use with your genesis characters, DAZ would have defeated their animation shortcomings and could easily fill their coffers from the ignorant amount of useful plugins and accessories that would come from the marketplace!
Came across this video on what is going on with the class action lawsuit, and it is quite interesting as had no idea that one of the folks behind Corridor Crew is a lawyer. The video is quite informative to say the least, and the outcome will be interesting. Near the end of the video he mentions that Getty Images has now filed their own lawsuit as well.
they will need to tread carefully with that lawsuit if they don't want to throw the baby out with the bath water.
Many budding artists start out doing fan art, they also scrape the internet to train themselves, if the law can be applied to a non human entity which works in an agnostic way it certainly can to living human beings making conscious choices.
Getty, Disney, Warner Bros etc almost certainly will incorporate ai technology sometime using their own legacy paid for content to train their databases and purchasing content from each other, it's the smaller players that will lose out here.
successful artists have already gotten paid, Greg Rutkowski got paid, Jane and Joe Blow sketching a furry fox avatar and uploading it to their Deviantart gallery hasn't seen much income yet and it fleetingly resembles a certain fox in a popular anime.
Again I am not suggesting ai trained on the LAION 5b database should be used commercially, on the contrary I would say it would need to at best be on a case to case basis on how transformative the product is at worst an ai image that closely resembles an existing artwork would contravene copyright so it's a use at your own risk tool.
But those laws already exist.
Interesting video. Implications of the court verdicts, if against using AI, would affect its other uses.
And that is the thing, AI does have its uses apart from making art. And very true they will have to tread carefully, as there maybe other implications such as someone creating a picture derived from another persons artwork.
By definition all AI Art is derivative. The court cases can't come soon enough. It is the big players like Disney and Pixar and Warner Bros. and Marvel that need to see the threat posed and bring their deep pockets into the fray to protect the little guy real artists before they find themselves in copyright jeopardy themselves.
https://en.wikipedia.org/wiki/Victor_Gruen
https://www.daz3d.com/suburban-shopping-mall
Yes, and after 17,000 shopping Malls sprouted up across all of North America and the rest of the world, I built by own 3D version of a shopping mall, from scratch, moving each and every polygon in the program called Hexagon3D by myself. I have each and every build version file along the process that I can verify in court if need be. No one owns a concept, but the tangible form of an idea is copyrightable. That is the protection that protect Intellectual property. That's the difference. What the AI compnies are doing is stealing the actual tangible forms of the artist's work which is protected by copyright.
And Wendy, I have no problem with you doing your thing. It is very creative and sometimes quite funny and inverntive. But you are not trying to benefit from it commercially at the expense of other artists.
I was not disputing that just an example of how most stuff derivative which is OK
I like and use your model a lot
my point is laws once made set a precident and one must tread lightly,
3D modellers may be the next target, they certainly have been already for some of their subjects hence the Editorial License being introduced here but soo far protected under fair use and transformative use
I am not pro ai either just don't want to fall off the wrong side of the fence
I am also making myself unpopular on both sides it seems, those who fully embrace ai dislike my criticisms as much as those against it, as one person pointed out, this debate has taken on a religious like fevour
You are using the term derivative inaccurately in this regard. Derivative, when it comes to copyright, is very specific. In architecture, A building is a concept, a shopping mall is concept, a tower is a concept. How that concept is interpretted and made into a real design is intellectual property.
It is only when the architect or the designer puts pen to paper and designs the structure does it acquire intellectual property protection.
So the Toronto CN tower and the Seattle Needle are both towers but each is unique and has design protection. Neither is a derivative of the other but they are both conceptually, towers.
After 17,000 shopping malls have been build they are conceptually similar but each architect's set of drawings, are unique. Again, not derivative. Just similar.
But when it comes to AI generated art, the very process of scraping and deconstructing the real art data set makes the resulting regurgitation a derivative of the original works.
AI generated art is derivative. And that is why the Copyright office is not granting copyright paperwork for it.
Does anyone follow the standards bodies / orgs that standardize the metadata tags used in digital photos and other digital image formats? Any movement there towards incorporating tags to grant or deny permission for use in Machine Language learning or AI image generation?
Fair use would be fun in the international context... fair use, but not for EU-customers of your network, for instance. Or get blocked, or 10% fine :). Stuff like that.
If you wanted to avoid such, the minimum measure simply is strict "opt-in" only, signalled via metadata. Should be in HTML, e.g. due to older formats. And re-training only after re-checking, so removed images can't be used anymore. Anything else would be nuts. I don't care how a certain court case ends here or there.
Also consider, that all sorts of formats including video will be scraped same way. The AAA(A) industry will always be ahead and make idk/ 2 billion in a day or what not, but the whole of Hollywood? Think they're having a glimpse around the corner already :p. There is no way to stay unique in any way, except if people let you. Perhaps this hits the storytellers hard, who try to stay within some cosmos. A not copyrightable thing still means, that you can't have copyright for it either. No copryight might still not prevent some platforms from providing monetization, and in return, demonitization for the original artists? It may get more complicated with series of images, like comics, though.
Edit: The artwork side really seems to remain a carousel of worms. It seems to be hard to escape from it. While law may be spoken, it's not certain that it will actually resolve enough issues for this to make sense.
Looks like everybody makes different kinds of reasonings and estimations, but still all seem to be off somewhere, and partly considerably off, concerning all the problems that follow. Maybe this is due to the fact, that the whole issue can't easily be seperated into different problems to tackle individually or sequentially. That's also where some of the failure potential with courts might lie, in case of a somewhat confined focus.
Concerning off or on, a current article from the law side, also containing a quick description of the algorithms, hints at possible consequences of a precedent. https://www.fieldfisher.com/en/insights/ai-related-lawsuits-how-the-stable-diffusion-case
It reads a little bit like "they could enforce software quality with future ai developments", or rather you could sue them for bias, which in my opinion is something different than using a training data set that threatens to poison the source of the data itself. Something that doesn't work or break at random, already is well known, e.g. modern operating systems. But violation of copyright (if) or the circular reference with training data of a system that relies on but at the same time reduces the training data by it's nature, in my opinion is a different category, because you don't have a new world with different or new jobs, or just a better but otherwise independently usable tool, and everybody moves on happily ever after. You might as well find yourself holding a can of worms in your hands, which likely won't go away just so. Do we tell the "neuron artist entity" [subtly differently pronounced ~ "moron"], that we take their stuff anyway, and the ai will be great, for the benefit of all mankind, who now all become clickwork-illustrators in a day. Undead artists, after all. In the end, using literally everything is just faster and cheaper, not necessarily better. But everyone being able to create images isn't the main problem, you can simply flood any site with quasi-copies of images and styles, once an artist publishes anything at all. Copiers won't be specialists anymore, but could literally be anyone. Resolving this by effectively removing any protection would be cheap, though you'd always have to consider what further protecting (which/what) means in terms of damage, otherwise. An interesting problem with the "everyone scenario", of course is where to get training data from after a while (can't repeat that often enough... will we just say "up to 2022... enough"? Also this means telling that the artist face-to-face now.). Resolving that needs more/different measures than enforcing either of fair-use or classic copyright for this case, the consequences are too far reaching. (In my not so humble extrapolations.)
(One more can: So deciding for fair use, will make it necessary to find a way to tell, what works actually are useful for society in particular, if you needed to filter or find a way to reward originality, being confronted with an "upscaled amount" of similar images. Will "best rated" be what we call "art" then? We might create what fair use was to prevent, in the end :p...)
https://www.wired.com/story/how-to-spot-generative-ai-art-according-to-artists.
I saw it as well. Very interesting.
True, so you have to supply it with images etc and train your AI seeds.