Mind Blown: Blender + AI via ComfyUI and Flux = 3D2IMG
Nyghtfall3D
Posts: 813
nVidia has released a set of AI Blueprints that can be used with Blender via ComfyUI to turn un-textured objects and scenes into photorealistic AI art. The caveat is that it only works with 4090, 5090, 4090 laptop, 4080, and 6000 series GPUs, and requires 48 GB RAM.
System requirements aside, this is what Daz Studio AI should've been, and now Blender has done it.
Son of... a @#$%.
Incidentally, I would dearly love to see folks who don't like like AI weigh in on this. I am genuinely curious to read your thoughts.
EDIT: Revised last line to remove inflammatory wording.
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
Have you used it? I thought about it, but I don't particularly like using Blender.
I don't have any of the GPUs it needs - I use a 3090. Otherwise, I would love to try it.
This is indeed the kind of "ai as a render engine" workflow I'm still hoping Daz or one of the PA's will put together as a tool we can use inside Daz Studio.
Why I don't think it's accessible to everybody yet:
NVidia is taking a huge step in the right direction by addressing the constant back-and-forth toggling between apps, setting up image inputs, etc. This should make the ai stuff more approachable to more artists. Blender has an edge here because both ComfyUI and Blender natively use the Python scripting language (Daz uses something else). It's cool to see stuff trending and this way and I'll be playing around with it for sure. :)
Its not about hating AI, its about not caring about it. I like to create, if the AI is doing the creating then that just takes the fun and satisfaction out of the creative process. It seems everyone that acts like AI is so great, really does't enjoy the creative process, they really just want the "upvotes" or "follows" and they are just looking for a means to an end. For me the journey is the reason. I just like to create. I don't care if anyone sees it or likes it. So to me Daz is fine as is, with the exception I would like better animation tools. But I make do. But I will doodle on a napkin in a restaurant or in the sand with my toe. Someone is either creative or they are not. I think that is the real divide.
+1
Doing things "the hard way" also develops general skills and abilites which benefits you in other contexts. Body and mind need to be kept active in order to perform well, physical and mental abilities are slowly deteriorating if not being used. Life is about growth and development, becoming more instead of less. Life started as tiny cells which hardly could do anything, it slowly evolved into humans, the greatest and most intelligent creators on the planet, because it wanted to become more. Technologies are great only if they support that desire for development in a meaningful way. It you want to develop your arm muscles, you do weight lifting. The weights themselves are a useful tool in that context, but using a forklift to lift them with is meaningless.
I cannot see anything particulary different to just using Canny in ControlNet with a render
it can be a render of primitives or a DAZ openGL render, I already do this with Fooocus using a 2080Ti
Well said! I totally agree. As a side note, I "love" how perspectives like yours and mine are reduced to "AI hating."
I don't even care if people don't like Ai but I am a bit shocked how many openly hate me and tell to my face because I sometimes use it
There are much worse things in the world to be passionately fired up about
NM see it was solicited
no matter how pretty you might think it is, it will always be a compromise and largely the result of someone else's work.
Nyghtfall3D,
I love it! Please share this news on the Blender Discussions forum thread.
And now, a few follow up remarks:
[1] I am cursing too ... because of the price I have to shell out to upgrade my NVidia card.
[2] Blender or Daz, AI or trad-3D ... I'm agnostic. Does it matter if the cat is black or white, so long as it catches mice?
[3] Lastly, for a forum that slaps down any talk of religion or politics, other forms of dispute gets a pass. "hate" and "love" (in bold) in the same breadth; the spectacle of people going after one another ... tsk, SMH. To be a wee bit dramatic here: this is the stuff Tennessee Williams, Victor Hugo or Dostoyevsky spun into tragic stories.
Cheers!
I've revised the last line of my original post to remove the inflammatory wording. Thanks for the feedback on that point. I was just so shocked to learn that nVidia and Blender both have developed tools that incorporate the same type of AI that Stability and Midjourney are being sued over for copyright infringement that I'm curious to know what, if any, impact it might have on positions held by people who loathe AI. I know a few artists who are quite passionate about their disdain for the tech - I used to be one of them.
I use it as a tool
Already got hate for using DAZ and Poser back in the day
I don't believe in commercially using unethically trained models though and only do it as a hobby without monetisation
I also use free options but am considering paying to animate my own renders
Cam Fox,
I tried to find a link to an interview with the NVidia CEO, the one where he declared that NVidia was focused on promoting the gen AI ecosystem by developing building blocks that other industry players could use. (Sorry, I can't find the article.) To this end, this new Blender+ComfyUI+Flux project appears to have moved NVidia's software stack forward, as well as stand in as a proof of concept that others could take inspiration from. My guess is that this will spur other DCC tool makers to adapt the feature set -- built on top of NVidia's framework, of course. It's pretty exciting to speculate when Daz or Adobe or Autodesk and others will bring similar functionalities to the market.
For now, as you've pointed out rightly, the workflow is anything but easy and intuitive. The value-added would mean streamlining the experience, hence lowering the barrier to widespread usage; plus delivering adjacent value in an asset marketplace. It would then be up to the users to pay more for all of this. In the meantime, I'm eager to see how people will discover interesting things to do in this new Blender sandbox.
Cheers!
AI has become something far more detrimental than what we're really discussing about here and in the other debate thread. We have people using it to bypass learning skills and critical thinking. Schools are actively fighting students using AI by ironically needing to use it to detect it. Society suffers as a result of this; a place where students don't learn how to solve problems and use their brains to analyze material. Learning something like calculus benefits more than the actual subject-the human mind benefits greatly from overcoming adversity and mastering skills. It's how we grow mentally, psychologically and artistically. If we stay on this course, it just leads to more laziness and people not really knowing how to do things. The emotional benefit of succeding at a difficult task is extremely underrated and I see people give up so easily these days, because there are just shorter paths with little effort. It's a bleak future.
For a lot of things in life, the time and effort spent to create something is part of the process and why something is special. Like the Sistine Chapel and Mona Lisa. I guess AI is good at one thing-preserving the beauty of non-AI works of art across all spectrums. I'd rather have one Michaelango's David than 10,000 AI generated images of customized dolls.
You seem to be under the misunderstanding that a person using AI isn't creating. It takes a lot of work, knowledge and some skill to get what you "actually" want out of AI. Most of my AI time is spent with AI inpainting which isn't any different than post work and that is more "creating" that loading some 3D models and hitting render IMO. DS is doing the creating for you, you are just controlling the input, which isn't any diffeent than AI doing the creating and you doing the inputing. I suggest that everyone at least try out AI before that make up their mind one way or the other.
There are huge differences between Daz Studio and an AI generator. DS requires the user to make conscious decisions about what they're using, what to move, what to actually tweak, while visually interpreting the (electronic) canvas in a full 3D space. You just don't do that with AI. There's not much of a creative process to AI when someone can search for prompts and constantly reroll their creations to hope for the best. I've used AI and there's no absolutely no sense of accomplish or satisfaction compared to anything I've done with DS. I have a lot of fun with my renders because I am actively manipulating everything, get my creative spark moments the longer I work on it, and enjoy how everything comes together. That's the human, creative process missing from AI. The render feature is nothing at all like a prompt. When you do your render, you're set on what you want to finalize. With AI, you just don't know what's going to happen. And you're always dependent on someone else's work, with no discipline or conscious decisions required to use it. With DS I am using everything intended to be used creatively, and purchased from an artist, too.
And to be clear, the aversion comes from mostly photo generation-postwork is something else altogether.
I have a pretty good idea on what is going to happen based off my experience and knowing the AI tools I am using, but we will agree to disagree.
we have been down this road before
same arguments different tool
and the irony, oh the irony
I too put far more work into stuff I do using Ai tools than I do my 3D renders using assets
and lets not even get started on the 3D mesh outputs from Meshy3D, Hunyuan, trellis etc
I spent more time in Zbrush, UltimateUnwrap3D snd Carrara fixing those blighters than I would have sculpting them from scratch
but
I never would have sculpted them in the first place as I would have never thought of creating what the Ai did myself
also the one made from a photo of me for example
I have morphed, face transfered and dressed the F out of countless generations of DAZ people to look like me
but that lowpoly badly UV mapped Ai generated prop looked more realistic in a scene than anything I could have done in D|S from a distance
of course I had to make it more detailed using Spotlight snd polypaint and it breaks up sundiviiding so remeshing was needed too
and you know what
3D modellers do exactly this using a device called a 3D scanner or a camera array and don't call that cheating
When it comes to creativity, The only critic you should really pay attention to is yourself. If AI fills a void for you go for it. If it seems like cheating to you it is, but only for you. LOL
The 4090 and the 5090 have a serious design flaw in the power connector and Nvidia does not want to fix it so if you want to thow away money to have a expensive but flawed AI video card and have deep pockets be my guest.
Saying that Daz Studio isn't creating is like saying a portrait photographer isn't an artist, because they didn't make the model, or the stool they're sat on, nor the lights they're using, or... etc, etc, etc.. (And admittedly, "photographers aren't artists" is something people have said various times in the past, but which I would personally consider to be bunk. Whenever you see someone from a photography background come into 3D, they *immediately* have a huge head start.)
The thing that makes me think less of generative images is that, regardless of how much skill or knowledge you have, ultimately you only ever get the result you want through chance. You could be the best prompt engineer in the world, but the process of AI generation remains "I'll generate twenty (or a hundred) versions, pick the best and refine from there", because ultimately you're handing most of the control over to a black-box random number generator.
You justified it by saying that it takes a lot of work to get what you actually want - but that's exactly why I dislike it. Isn't it supposed be making things more convenient?
But no, it makes you take effort because it can't follow simple instructions like "five women standing in a straight line. No, not six women, no, not hugging each other, no, not four women either, no really, I mean it, they're in a straight line and absolutely not hugging". At least with Daz Studio I don't have to spend time arguing with it reinterpreting my instructions however it wants, and I can, if required, explain things in painstaking detail by going to Blender to make things from scratch. (Whereas, with AI, I can feed it an image that's almost exactly what I want, just change the style slightly, no what are you doing, that character now looks completely different and where have her glasses gone?)
Ultimately, I think the process of AI image generation is far more analogous to commissioning an artist than being one. I mean, that's describing the picture you wanted and then asking for edits until you're happy... Admittedly, the commision artist in question is very drunk, barely understands English, can scarcely draw hands, keeps trying to work in their own biases, and is constantly tracing other people's work, but wrangling their incompetence doesn't make you personally into an artist.
But here's a thought experiment. If it taking "work, skill and knowledge" is what makes it art, does that mean it will be less art as the technology improves?
This sort of thing was already doable via ControlNet inputs; I've tried it there, but the reality is that if you give it vague inputs, you get vague outputs. Poor control over things like lighting direction, the thing making up where it thinks shadows should sit, the like.
I can't try it directly, because apparently for whatever reason they haven't made it accessible to the 4070 Ti Super, which only differs from the 4080 because it's a lower binned chip with a few less of the cores on the GPU die activated (but it's the same 16 GBs of VRAM, the same GPU die other than that, etc).
Looking at Blender official website, there is no news on the subject.
Reading Nvidia website and following the link about 'ComfyUI-BlenderAI-node', I ended on a Github repo that is not related to Blender (here the list of the repos managed by Blender) but to 'A non-profit open source organization dedicated to Blender and other interesting open source development work' whose websites are both in Chinese or Japanese (I can't read any of these langages).
So saying 'now Blender has done it' feels like a misrepresentation of what's happening when what's happened is that someone unrelated to the Blender Foundation has done it using Blender.
Agree COMPLETLY ^
Also I see no evidence in the video that
this system can create consistent restyled animation frames from my Blender scenes.
And finally for me ,and many others, the uber high hardware requirements and local complex setups defeat the whole purpose of using AI in the first place.
I can literally draw simple sketches on my
Samsung galaxy S24 Ultra phone and the built-in AI turns them into images, in a variety of Artistic styles, in seconds using googles connected AI on android.
I agree, especially with the highighted points. The first one drives home my first post about the way society is going, too. Ironically, sometimes more time is taken to reroll and edit prompts to generate an AI image instead of taking the time to actually learn a practical skill. So many people aren't learning how to actually do things.
My own proficiency in DS has allowed me to make images not only faster, but actually fit my vision, and are the result of me learning the program, artistic techniques and how to go through tools that properly fit my (or someone else's) vision. I was unhappy with the lighting in my scenes so I took the time to improve it-and I am really happy with the results. I think people forget that part of the artistic process is problem solving and the journey we take to be the best we can be. The critical thinking and utilization of resources to understand and aquire a new perspective is also important. This is what is missing so much with AI (across all forms of art). AI in most cases is not being used as a tool, but as a replacement for the actual process itself.
Secondly, one of the best feelings is when a customer commissions you for something and you're able to put their vision to life. The time, effort and empathy used to do this is what matters the most to me. Plus, feedback is important to further help improve my own skills for the future.
I don't dislike AI per se, aside from unethical training issues. I just don't want to use it. I prefer to use Daz as it feels more 'hands-on' to me. I see what I'm creating and, when it's close enough to what I imagined, I hit Render.
I don't get that feeling from using AI. Perhaps if I had better AI knowledge and skills that would be different. But I don't think so.
As for those people who use AI, I think 'good for them'. I certainly don't see it as cheating. It's just a different way of doing things, with different inputs and different outputs (in that you don't have such control and you don't see what you're going to get until you get it).