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
To be fair, though, you could probably dig up similar quotes from art critics in the 19th century decrying photography.
I think critique on (uses of) ai is pretty well informed these days. Not everything everywhere, but a lot.
To be overly fair, you could add-in how photography has lead to filtering and censoring technology, as well as surveillance all over the place, and that only for the sake of copyright, not even considering further implications of general mass surveillance :p. People often complain for the wrong reasons...
I think automation & AI will ultimately wreste control of knowlege and power from ego, even those at the current pinnacle of wealth and power, and actually problem solve for harmlessly co-existance altruistically. Of course that is a huge blow to every ego, but after all our brains are just limited mathematical solution sets; and those solutions that cause harm or don't stop harm need to be filtered out and discarded, and not used to advantage one individual over another.
That said, the Harvard Ivory Tower style answers chatGPT ridiculously gives currently are all to closely parroted after well established and unflattering existing ego driven human behavior, so I have hesitancy to fully endorse, and am hedging my non-bets.
there is an interesting image to 3D generator which OpenAI is collaborating on
https://huggingface.co/spaces/stabilityai/TripoSR
not UV mapped had vertex colours so unwrapped it in Zbrush
video showing more
It's been a while since I created any new 3D content. I still like remixing old product in Stable Diffusion and think that a well executed plugin or a render engine that allowed the user to incorporate diffusion into the rendering process in real time (so they wouldn't have to take the image into post work in another app) would allow users to have an amazing amount of control and create breath taking images in a variety of styles that would transcend the usual Daz/Poser look.
I think the AI is causing me to come back DAZ Studio more frequently lately. Because the AI adds the realism missing from DAZ Studio, and it is somewhat exciting to see all of the DAZ scenes look so much more realistic.
That's great work Joe, it really enhances the 3D render without changing it too much (except for the toony one, which is also excellent). Can I ask a little about your settings? I end up using Controlnets with LoRAs and all kinds of complicated methods, but it looks like yours are more simple and clean.
Below was the first pass on the zombie image. I then went back and did in painting element by element until I got what I wanted. Some things, like the police lights, never quite got there.
These were my settings.
I've been using Pony models in Automatic1111 still. Never been super comfortable with node based interfaces.
I tend to use a linework or canny control net with sometimes secondary depth one.
There are things about the feathers I liked better about the earlier pass. The bigger irises also looked less AI, but the smaller ones communicated fear better?
source_cartoon, rating_safe, score_9, score_8_up, score_7_up, score_6_up, Scared fluffy white male duck in brown plaid suit jacket (white shirt, red tie) standing on police car, orange duck legs.
Negative prompt: makeup, eyelashes
Steps: 30, Sampler: DPM++ 2S a, Schedule type: Karras, CFG scale: 7, Seed: 3984617828, Size: 1154x1500, Model hash: 50598d2be2, Model: lucidCitrusPony_kiwi, Denoising strength: 0.75, ControlNet 0: "Module: depth_anything_v2, Model: diffusion_pytorch_model_promax [9460e4db], Weight: 1.0, Resize Mode: Crop and Resize, Processor Res: 512, Threshold A: 0.5, Threshold B: 0.5, Guidance Start: 0.0, Guidance End: 1.0, Pixel Perfect: False, Control Mode: Balanced", ControlNet 1: "Module: lineart_realistic, Model: diffusion_pytorch_model_promax [9460e4db], Weight: 1.0, Resize Mode: Crop and Resize, Processor Res: 512, Threshold A: 0.5, Threshold B: 0.5, Guidance Start: 0.0, Guidance End: 1.0, Pixel Perfect: True, Control Mode: My prompt is more important", Mask blur: 4, Inpaint area: Only masked, Masked area padding: 32, Version: v1.10.1
Thank you for that information and especially the raw image. That's a good example of how it's not as easy to make AI images exactly as we want, as you obviously did a lot of fixing to get the final result. I hate nodes too, but I did manage to learn enough Comfy to make my own decent workflow. So you basically just put the DAZ Studio render into A1111 with depth and controlnet (I do that too, it does seem to be the best combination for these sorts of images, ), then generate until you get enough good bits and pieces to postwork together? That's pretty much what I do. Anyway, thank you very much for the information.
Using AI for improving a render seems to be a good idea. As long as the AI is still controlled by the artist. Otherwise the AI is the creator.
I really like the result by @JoeQuick
I'm going to post a few more, as these things have mainly just been sitting around my computer with no where to go.
Middle is the original.
So the image at the top is a DAZ render of what I think are three fairly realistic looking characters. The image in the middle is the same three with the AI set to medium. The image at the bottom is the same three with the AI set higher.
I find it useful for re- styling images.
I used low quality preview renders from blender
and re- styled them to create a 92 page graphic novel.
Do you feel like they're still the same character? Does it matter?
Middle feels the most like your original. Last one feels like the models idea of beauty is taking over.
Animotions? Any connection to the old super hero content centric Poser content site?
I agree, the middle one of the three looks like it's just been improved enough to still look like your original. Good job Wolf, the characters look nice! You kinda lost everything to do with the background, but if you just wanted a generic sci-fi room, it worked well.
Thanks Joe, I think the third from the left looks the best. Very realistic.
Maybe I'm only speaking for myself here, but DAZ - THIS is what we wanted from a DAZ AI. AI that improves our own renders, not creates them from scratch for us.
I do feel like the bottom image changed the characters too much to be the same people I imagined. I think I generally prefer the medium AI creativity over the higher creativity levels. But usually I end up combining elements from various pieces. For instance, in this image the hair from the bottom picture is applied to the characters in the middle picture. The AI is better than DAZ at hair. And sometimes I leave in things from the original DAZ image. Sometimes the AI messes up the hands and feet and DAZ gets it right. And the AI I use seems to favor positive facial expressions over negative so I have to incorporate some of the DAZ facial expression if I want the image to express negative emotions.
As for how much it matters to me if the characters aren't the same as the original, sometimes it doesn't. Because often I'm trying to depict a certain type of situation in the scene. So it doesn't matter who the characters are as long as they fit the type of situation in the scene. But I do have my favorite characters both DAZ and AI, and I wish it were possible to reproduce those AI characters consistently.
my impressions:
I remember when I first stumbled on this 3d hobby, long before I ever learned to model the kinds of content I eventually made, I came across sites like Animotions and Heromorph. Homes to crude but effective superhero props, often items made of distorted piles of primitives. We kept pushing things farther and farther at Daz, but now we're back to a point where much of that kind of content, just with a layer of diffusion over it, would do just fine.
In addition to being fond of thse "protuberances" (best description I've read in a while), the AI also seemed determined to take away what I think are supposed to be shocked or startled expressions and turn them sultry.
Thanks for sharing the images.
I think AI art is far superior to any art ever created by any human. AI is explosively exciting. You can make AI as personal an expression of your imagination and subconscious as any art medium that exists. Art museums are mausoleum of dead art that we've been looking at forever. It's time for something new. Just like it was time for something new when Poser and DAZ Studio became available.
Until people start using it like we are here. It fosters anxiety and panic because like Fauvist says, it has the potential to be better than any human artist. If we use it as a tool to enhance our own work, much of the creativity and even the mistakes can still remain.
It fosters anxiety and panic because it robs every artist while trying to destroy the ability for artists to do work.
The glimmer of hope is that it is massively self destructive.
Between encouraging artists to spread poison works, discouraging public art (which then limits the theft that drives genai), and ai artwork poisoning itself (much like careless organisms spreading feces in its own food source), thankfully it's likely to kill itself before too long.
Maybe once that happens we'll actually see ethical genai.
AI isn't killing it. Especaially if you'd seen what DeviantArt's DreampUp generator just came up with.
There's a suggestion doing the rounds that AI is becoming in-bred. If that's the case, it's not going to improve. It's going to stagnate (like a lot of people said it would a couple of years ago).
I've got ChatGPT helping me out with a new project, but it's mainly for advice and suggestions (help with coding - Because I'll need it ^^' ). I've also been using Suno to create songs, with lyrics from ChatGPT.
As for my own art, I've only ever used to to remix some drawings. There's only one (out of something like five) that look better than my originals. I've also installed an AI program which can take low quality images and rework them into 4K+. But I can't see the improvement in the pictures.
I don't see AI doing anything that someone can do themselves if they tried. It would actually impart some of the artist's own human experience and not just rely on someone else's definitions and opinions.
We will never see ethical AI as long as it is open source and allows anyone to train models. Everyone wants to blame corporations for the mass scraping of public images, but the chance that anyone will ever generate an AI picture that contains a recognizable portion of one of those images is miniscule. The problem lies with the individual users that train a LoRA on a dozen images of a specific artist's style; that is when you can get generations that look too much like the original artist's work (not to mention most individually trained models are trained for and on pornography. The commercial models are most certainly not, I cannot even get them to generate the PG-13 stuff I want to make). It is an ethical dilemma because while I don't want traditional artists to suffer because of AI, I also think everyone should be able to have the opportunity to illustrate what they imagine.
I don't think it's going away, I think artists are going to need to adapt.
Virtually every time I run one of my more simple renders through AI, it is a massive improvement, especally in hair and clothing. If AI were to go away, I would really consider just not making art any more because I'm bored with the same 3D look after 25 years and making 2D art isn't worth the headache when I'd be lucky to get 200 views on it.
edit: Heh, when a 3D program can render hair like this, have it float on water like this, and be able to render in a toony style that doesn't still look like CG from the 80s, I will delete every trace of AI in my system.
I watch stuff like this on Youtube
https://www.youtube.com/@zythi
https://www.youtube.com/@jackvideoai
https://www.youtube.com/@OscuroSeptember-AI-Channel
https://www.youtube.com/@iaskedaitodoit
https://www.youtube.com/@MeanOrangeCat
https://www.youtube.com/@YunasWorldsAway
we are never going back
I could say something similar about AI making me come back to DS, but with a very different and inverse reason.
I muck around with AI image generation on and off (although seldom share the results), as I find the underlying technology intriguing, and it's interesting to see what it can do. However, I ultimately always find it limiting. Some things it needs massive direction to manage (like sketching in stuff and completely re-engineering the prompt to inpaint that area), other things it can't manage at all. Expression and eye control is frequently abysmal on many models (particularly if you absolutely do not want the characters looking into the camera), framing is frequently out of your control, etc, etc.
Fundamentally, it does not understand what it is being told to depict. It can imitate what it's been trained on, but what it's been trained on is a highly skewed data set. Portaits looking into the camera? Lots of those in the data set, so not too bad. Rear three-quarter views of someone's head? Expect alien monsters. Hands in weird or complicated poses? Hahahahahaha.
So for me, I've often had an idea when playing with AI, try to do it there, then waste enough time and get frustrated enough that I just fire up DS and do it there instead. Because there I have full control and I can make definite progress towards the result I want, rather than trying to explain it to a robot commission artist who doesn't entirely understand English, thinks they know better than me, and can't do hands.
My efforts to AI process my renders also usually lead to disappointment. The one area I've had any real success with it is using it to inpaint hair. Trying to use it for faces is just a recipe for inconsistent characters, sultry expression and eyes not looking where you want.
It's technologically fascinating, but every time I try to use it for something more in-depth than low-effort "show me Elmo as President of the United States" trash, I either end up disappointed or invest far too much time in getting something mediocre.
AI creates some pretty good art at times, and can be creative in unexpected ways combinatorically that people never think to go to, although most of the time those strange combinatorics are simply a mess.