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
As for AI ripping off other artist's copyrighted work - there is a million times the amount of visual images that are public domain than are copyrighted. 9000 years of visual art is not copyrighted. Millions of paintings and drawings and sculpture from every country and every culture, and from every era in every style. AI is taking the faces, and bodies, and the poses, and the expressions on the faces, and every aspect of every piece of clothing, and every setting and location, and every mood, and every atmosphere, and turning it all into data that can be recomined in unlimited combinations. Every photograph produced by the American government is public domain. Even photographs they produced today are free to use. How many million of those are ther? Photogrraphs that cover virtually everything that happened in the 20th century. Tens of millions of photographs and films that were never under copyright because they were produced in Communist countries. A hundred million images where the copyright was never renewed and so the copyright expired throwing 3/4 of the commercial images from the 20th century into the public domain. At 24 frames per second, there are an average of 129,600 individual photographs in a 90 minute feature film. Multiply that by how many feature films, and short films never had their copyrights renewed. Virtually every pose a human body could possible be in is in there somewhere, every shade of every emotion, all free to use.
But then there are artists who have NONE of their visual art in the public domain. Name brand artists who are famous, and not famous. Artist's who are still alive and still working. Can AI create an Andy Warhol painting without using any of Andy Warhol's visual art? Just by typing the name Andy Warhol, can AI "figure it out on it's own" and shuffle all the data to put together an "Andy Warhol" painting without actually ever using any Andy Warhol painting? Is there enough periphrial material surrounding Andy Warhol that AI can "imagine" what an Andy Warhol painting would look like? i remind you that Andy Warhol became famous for taking copyrighted, tradmarked, images and "appropriating" them for use in his silk screen canvases. He may have received permission "after the fact" when those images became more famous than the source images. But did Campbell's Soup and Marilyn Monroe's Estate, and Elizabeth Taylor, and Jacquiline Kennedy Onassis sign away the righfs to their images before Warhol created the canvases? No.
I think a computer could create every face that ever existed, just by using the images of Jesus Christ that exist. By rearranging everything. There's two eyes, a nose, a mouth, a head, hair. How many different images exist of Jesus? Make the nose a little thinner, the eyes a little bigger, the chin a little more female-like, the hair a little more curly, the complexion a little darker or lighter - a face is a face. Can Ai use just that one face to make every face? Sounds like morphing a Genesis figure to look the way you want.
That has amazing potential, unfortunately it is not what is actually happening. In my experiences at least, you have virtually no fine control over how an AI face looks unless you use a celebrity LoRA. I've tried prompts like "larger nose", "round face", or "strong chin" and they rarely ever have any effect. Some models are trained so heavily with images of Asian women that someone made a LoRA to actually "de-Asian" the results in order to have a chance at getting someone of another race. I have one model that gives great results, but it always tries to turn men into women. Almost all AI generated people that are not the result of a character or celebrity LoRA have very similar faces, and it is one element where 3D or manual creation are still far superior to AI.
One example where I have some experience is trying to generate black women with afro hairstyles. Almost EVERY ONE, no matter the model, will include the woman wearing hoop earrings. It makes me wonder if only a couple of images of black women were scraped in the intial creation of the original models. I used a Frazetta LoRA to generate a woman with an afro wearing 15th century clothing and it still put gaudy hoop earrings on. When they talk about these models being trained on 11 million or even a billion images, I really wonder what those images are actually of.
I wonder if someone could take something like 128 photos or so of the human head from very slightly different angles, then train a LoRA with them and assign a numerical code to each angle. Run that alongside a model to try and force the checkpoint to pose the head in the position based on the code you include. If that were to work (and I have no idea if it would), it would be far, far more valuable to the community than another ball gag or sagging breast LoRA. (I know you can pose the head using Controlnet, but it's still finicky, especially with newer models).
As I understand it AI works by generating a cloud of probbailities for particular pixel colour distributions associated with key words. When you give it a prompt it takes a random pattern of pixels (generated from the seed value using a pseudo-random process) and then progressively adjusts the distribution of values (over the chosen number of iterations) to match those that go with the prompt words. It does not in any way analyse the input images into faces, clothes, cans etc.
US copyright law has always been a bit odd/eccentic, and I suspect copyright renewal is one of its ecentricities - though it is true thast laws across the globe have been updated over the twentieth century to try to iron out some of the kinks. In any event, whether the image is protected is secondary to the question of whether it has even been digitised and made available in a form that AI bots can scrape.
I use Midjourney. You can upload an example image, or you can generate images using prompt words. Not all the faces are the same. If you want a Goth character, or a Cowboy character, or a supermodel character - using those words will give you the type of face your looking for. Then, when you get it, you can hit RERUN, or redo slightly or redo a lot - to give you as many Cowboy or Priest faces as you can stand to look at. Do you use midjourney? If you don't, tell me what you want, and I'll see what I can generate?
This was my prompt: photograph woman african afro hairstyle
Sorry, I should have said I was talking about Stable Diffusion and the public models. Midjourney's have their own style by default, and I used to use it (and liked it, until it became too expensive and censored to get anything I wanted). Thank you for the examples - still has trouble making a nice round afro though? Haha, so does every other model I've used except some wacky toony Japanese LoRA. ;)
It is already there. this took five minutes in FooocusAI (stand alone on your machine) with inpainting. The first image is a snip from a render I did last year of a blond hitchiker . I loaded it into the inpainting tab, under improve detail. Selected the bag, typed "detailed bag", and got this on the first try. Then loaded that generated image and selected the oufit and typed "detailed outfit, tight tan crop top, shirt denim shorts, tall cowboy boots, black leather jacket, tan skin" and got what you see on the second image (the first one has the cleeves too short). Did the same for the face and hair, total time, just under 5 minutes. Granted there are some small things i would fix if this was a serious image, but for 5 minutes the results are not bad.
I was an AI skeptic for a long time until I decided to dig deeper into the tech and now it is all I use. DS is only used for posing since you can load images as prompts. If DS want to keep their market alive, they would do well to have some kind of inpainting took in DS.
I've just accepted the idea that everythig in the world has been scanned and entered into data bases - whether it's copywright protected, trademarked, top-secret, undiscovered by humans. I think it's all been scanned. I don't know if you remember, but 10 or 20 years ago Google announced they were going to scan every book in existance and put it on the Internet. If you didn't want your book scanned and posted, you had to opt out by a particular date, otherwise you were giving implicit permission for Google to take your work and scan and post it.
Many of the great Art Museums and other kinds of Museums of the world have uploaded professionally imaged copies of all their holdings and made them free for the public to take and do with whatever they want. The greatest museums and the greatest art collections in the world. All of it. It's mind-boggling.
https://www.rijksmuseum.nl/nl?gad_source=1&gclid=EAIaIQobChMIi9apot3iigMVD3RHAR3MkR7tEAAYASAAEgIi_vD_BwE
https://www.metmuseum.org/about-the-met/policies-and-documents/open-access
https://www.openculture.com/2021/03/the-louvres-entire-collection-goes-online.html
https://www.smithsonianmag.com/smart-news/british-museum-makes-19-million-images-free-download-180974799/
https://www.nga.gov/open-access-images.html
The Smithsonian https://www.si.edu/openaccess
The United States Library of Congress https://www.loc.gov/free-to-use/
But that's basically two different females... and it still works by the "give commands and pray that the outcome is kinda like what you want", what might work when doing single pictures, but not when going for storymode...
There are several ways to trick the AI into repeating the same characters....sort of
Depending on the style you are using
I am more interested in NPR/ stylized AI imagery
So I used very low quality Blender viewport snaps
to guide an AI for my camera shots to create this toon style for my graphic novel
as you can see I was able to re-create the antagonists leader for my story for the entire 92 page novel .
One thing you can do is make a LoRA of your Daz character. I've done that a couple of times, and it works out pretty good.
Basically, whip up 10 or so renders of your character that you want. These should all be varied in pose, emotion, clothing, but the one thing that remains constant is the face structure. What I do is set it up with 3-5 headshots: one front, one each 45 degrees, and if needed, one each of 90 degree profile. Neutral expression. Then a few "action" shots in various clothes, and poses making sure their varied enough that it knows what your character looks like with a dress, or jeans, or whatever style you're aiming for. Vary the lighting also. Again, just setup a bunch of quick renders to represent what you're after. There's a program (I forget what it's called off the top of my head, and my computer is down currently) that you feed all your images into (they just need to be 1024x1024), and you describe each one in a txt file the way you would write the prompt for that particular image. Make sure that you tag something the same in it also, like "Jessica, a 25 year old woman, she has long brown hair and dark brown eyes, looking directly at the camera, headshot". "Jessica, a 25 year old woman, she has long brown hair and dark brown eyes, profile headshot" "Jessica, a 25 year old woman, wearing jeans, white hightops, and a leather jacket, she's standing on a street at night, her brown hair tied up in a neat bun". The AI will learn that the "face" is "Jessica", so when you attach the Lora to your SD model and say, "Jessica is standing at a bus stop in the rain, her long brown hair is tangled and drenched from the rain"
Friendly warning I don't think you are allowed to make Lora out of Daz's content. terms or services, EULA etc... ( the Content License expressly excludes the use, incorporation, or input of any Content, in whole or in part, in connection with i) any AI engine, program, or system (including, without limitation, Image.ai, Nightcafe, Artbreeder, chatGPT, Shutterstock, DALL-E 2, Deep Dream Generator, Hotpot ai, DeepAI) with capabilities or instructions to auto-generate materials that are derivative, imitative, or otherwise plagiaristic of the Content; and ii) any activity contrary to the intended use of the Content License, or otherwise to circumvent the terms, restrictions, or safeguards applicable to this Content License. For the purposes of this EULA, "AI" means any method of artificial intelligence such as deep learning, neural networks, or any other similar technologies intended to consume and analyze content for the intent of auto-generating new content.)
ps: I wouldn't do public Loras only private / local on PC.
I think we're nearing the point where we're going to see if AI as a "make my art and make my fiction for me" means will be the mainstream go-to or if it's just a passing trend that will soon be relegated to niche dedicated communities. We've seen shiny new things in the tech and entertainment world(s) that ended up being just a passing trend rather than the Next Big Thing (that would stick around until the next Next Big Thing arrived to make the masses forget about that).
3D I think will still have a place either way, though adjustments and adaptations will need to be made. Painters didn't go away after the introduction of the camera, 2D sprite artists didn't go away with the dawn of 3D models...
Why do you think it's the future ? it just started, there are so many ways this can go down the drain very fast.
Facebook wants you to talk to "virtual AI profiles" ? is that adding anything useful to the world ? (the answer is no and it look like FB figured that out also now)
Most AI starting to get restricted, they block them from comment on certain names or events because some people will be upset or people have requested not be named by an AI, so in reality you cannot trust an AI, it may also lie or misinterpret information, so no matter what you do, an AI has to be supervised in most situations, many AI's always answer, if they don't know the answer they will simply make one up..
Youtube is flooded with AI generated crap, usually Sci-Fi "movies" with woman with very little cloth on them, yes they have lots of views but there is no real content, it's very boring to watch, it will die out soon.
AI can be used for many good things, no question about that, but so far we have seen very little of that.
And sooner or later there has to be someone that will deal with the criminal AI companies that steal copyrighted information and data they are not allowed to use for AI training, as long as you do not know what the AI is trained on you really should not support them (my personal view)..
And what good is an AI that cannot talk about many things because it's not allowed to, lies about things, invent things that never happend and then come China with an AI that cannot say anything bad about China but are happy to lie about other countries, it is happening now.
When you don't know what an AI has been trained on, you cannot trust what it say, because the AI has been tweaked to say what the company that created it has decided what to say.
And there we are again, no matter what you use an AI for you can never trust it, you have to check what it say is actually true, and what use is an AI when the information you get from it cannot be trusted and has to be verified.
When it comes to art I guess it is a matter of taste, but telling an AI to create an image, painting, book or video for you, why would you do that ? yes, you can massproduce "something", but it is not you that created it, why not do it yourself and create something that is made by you and not by (sometimes) stolen data from other creators, you would expect people to have some will to actually create something themself, I am not sure what is wrong with people these days, I have seen people complain about features in graphics apps, "But there is no AI so I can tell it to make a drawing for me".
And then you have YT creators like, oh, now I can make 10 videos every day, I don't need to say anything, the AI can create all that for me faking my voice, sure, from an income point of view I guess that sounds cool, but from a creative view ?
Yes, they add AI to everything, and in some cases it is cool, but again, it's a new thing, this will settle after a while, most things are starting to be sloooooow because of all the AI junk they put in there, Windows 11 is already on the way, and they (MS) do the best they can to force you to use AI, you should not be able to disable it.
We use Copilot to summarize meetings as a nice bullet list, but that is not going well because it has started to lie and add things to the summary that was never said in the meetings.
We have nothing to worry about from any AI, but we should worry about the companies behind the AI's
It's of course up to each and everyone, but I take a stick figure with a good story any day over beautiful perfect art with a rubbish story.
10 years ago it was "Self driving cars are the future, in a few years we will not bedriving cars any more...", that is going well
Look up the story about the AI with the medical images and the ruler, the AI is what you feed it, and look at who are feeding the AI's.
Very good take on this. Personally, I don't think AI will replace actual artists. The humanity behind everything-the creative mind, the passion, the love and care that goes into making every line, stroking every brush, shaping every model and creating every texture, is the driving force behind exactly why art is so beloved. And, ironically, if there weren't any more artists creating anything, AI would only be more limited on what it can do.
I think AI needs to be highly regulated, simply for the issues it causes by, essentially, violating copyright and if not that, disrespecting the original artist. I'll also say it does has it uses, but I think it should be limited to the artist utilizing what they have created, mostly for enhancements or touch-ups. AI can be a great time saver for many things, but it has to come with one big thing: respect.
Bingo. True Art is with HUMAN EMOTION! AI can't give you that.
So in the first picture, on the left is what I think is a hand drawn image maybe done on one of those Wacom tablets. And on the right is what came out of the AI. It looks like the AI struggled with some of the cartoon like proportions of the drawing, and went completely haywire in a few places such as the feet. But I still think it's pretty interesting how much it could do with even a hand drawn image.
And as shown below, after cropping just the faces, I think I was still able to get a usable image from the whole fiasco.
In every work of art, the artist himself is present. - Christian Morgenstern
I just recently looked through AI stuff. This 3D star is not in Danger yet.
Quote:
"In every work of art, the artist himself is present. - Christian Morgenstern "
The problem is, in AI work, there is not much of a "himself" inside.
Two females and a total of... what... eight? hands... a double elbow, too... I didn't dare to count the fingers, which often are a give-away for AI going artsy...
Yes, that was my point.
I also notice that it turned a wrestling match into a sultry photoshoot.
Considering what most AI art is done for, that should be expected, when the AI tries to guess the user's expectations and does it's best to satisfy them.
You are probably right, no matter that prompts I used, there is probably some naked woman sommewhere in the castle having sex since that is all AI is good for.
I dunno - that castle looks pretty sultry to me, LOL.
Having said that - 'most' can be hard to gauge as we don't see most people's outputs. If there were a figure stood in the water in the foreground, would the AI try to make it a sexy photoshoot, or would it, without hours of trial and error, understand 'fierce woman in half-plate armour fighting off crocodile with a burning torch'? (I now have images of a torch-wielding crocodile but that's maybe just my brain).
This is using your example "fierce woman in half-plate armour fighting off crocodile with a burning torch" as the prompt (Krita AI). It needs more guidance, but I think it sort of works. The second image is one I generated to use as a background for zoom.
I think the theory that AI is only capable of producing sultry scantily clad images of women has been aptly disproved. If not I'm sure more examples can be posted.
The Passenger / Stella Maris piece and the Munchausen Wood Figures are ones I actually like. But the Captain Hook (even without a hook) and the Metro Station piece have their merits.
I don't know. Would you have thought they were AI if you didn't already know?
I get the Midjourney Magazine. I leave them in my classroom. Students flip through, not really knowing what it is at the start. But they flip. They seldom linger. I wonder if they would linger on real human art?
Valid point. I say most based off how much time i spend going thru the CivitAI and other AI image galleries. While there is a lot of adult type images posted, there are many that are not. The same can be said for DS, while NSFW images are not allowed here, that is what many use DS for.
I have some
good fun messing with AI once in a while. Here is just threequick and dirty generations cuz it's late. I have gotten some really cool painting ideas from generating random stuff in AI.
I never claimed that AI can produce "only" such pictures. Just that it seems to need extra effort to keep AI from doing it