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
https://www.theguardian.com/technology/2023/mar/16/the-stupidity-of-ai-artificial-intelligence-dall-e-chatgpt
You are exactly correct on this point. Going to dust off the old surfboard and catch that wave.
I am so fasinated with all of your AI work!!
I love working with my Daz characters, but I really don't like dealing with backgrounds AND I want them more interesting.
I would love to know an easy to learn AI program that I don't have to purchase a new computer for. I can pay for software, however not a lot!!
I would love to know some of the verbage you use to create your wonderful art.
I'm asking a lot, I know.
MidJourney uses Discord on their own hardware so you can get GREAT results on your phone, tablet, or any ol' computer! No real user hardware specs required! I use MidJourney to create complete images, or as backgrounds for renders. I also use it to composite elements into 3D Renders, and use 3D rendered elements to composite into AI imagery. They have different tiers that you can join in (currently ranging from $10 - $30 a month last I checked). The prompts are publicly available on Discord and there are a fair number of FaceBook groups dedicated to sharing prompts. There are websites with dedicated prompt generators where you make selections like Photo or Cartoon? Landscape or Architecture, etc. You can be as creatively involved, or as unplugged from the creative aspect as your needs require.
Create a Discord account if you don't have one, then go to the MidJourney bot and get started with a handful of free credits. That should let you decide how deep you want to invest.
Good luck!
[Edit] Also note, you are likely going to need to get some sort of enlargement software. A.I. solutions that run on your local machine are HIGHLY (and HIGH SPEC) hardware dependent. MidJourney just last night increased the output size to, I think 1024 from previous 512. (If my data is incorrect, please, anyone, feel free to correct me - I am at work and "keeping it light" here <LOL>) I havea couple of recommendations on the enlargement software and will post them after I get home tonight.
https://www.futurepedia.io/
https://pharmapsychotic.com/tools.html
something for everyone here
I have https://nmkd.itch.io/t2i-gui on my own computer as well as https://softology.pro/tutorials/tensorflow/tensorflow.htm
the first one is probably the simplest option for your use, it will download all the models and dependencies for you
just click the exe in the self contained folder (don't install it into programs just put the folder in the top level of your C drive is best)
it will take a while to download all the python etc stuff and install so be patient
Guys!!! Thank you so much!! I am so excited to give it a try!
I understand a lot of people disagree with using AI.
If someone wants to take one of my gallery images and change it for their own use, have at it!!!
In my opinion, this is the most correct thing said here, ever. When you think about it, we've had technological innovation for at least 20,000 years and yet we still have an economy, industry, and jobs. But a different economy, industry, and jobs.
I'm one of those not in the AI camp. It feels like cheating to me. Like using other people's work. I've seen some beautiful stuff created with it, yes, but if I was using it (I'm not) I would never feel like it was my artwork. It's nothing but an amalgam of someone else's work, and while you can say that Daz Studio is the same thing (and that would be partly true), I can still make my own stuff for Daz Studio (and I do) and I am the one who decides where the image goes, not a computer ;). Half the time, I don't even feel like I can take credit for a Daz Studio image, but much moreso than I could if a computer created the entire thing. Hope that makes sense. LOL I just don't find any kind of challenge in AI imagery and without challenge, I don't find it worth doing.
Artists will still be able to use AI as bases for art, but AI could really put print models out of work! The new V5 version of Midjourney is crazy! Even hands are much better!
Weird, this image got removed. Anyway I'm reposting to show how realistic it is!
I used some key words that are in my own art and I'm pretty sure these are my watermarks based on font and how they are positioned but it looks NOTHING like my art and is way too photorealistic but it makes me less concerned about our art being stolen or stealing art from others.
Surely there are artists out there that would be willing for their work to be used in AI for others. I understand about the money issue of it.
I think it's all going to come down to a huge copyright issue, but again, I believe there are artists, who are very good, willing to sell their art to AI software companies!
Yes, it means everything is more expensive.
I totaly believe in AI art!!!
Before jumping into midjourney with your wallet, I recommend experimenting at NightCafe. They have multiple versions of stable diffusion, Dalle e, and some clip guidance extensions to stable including specialty AI for inpainting and photo guided work. you get 5 credits a day for signing in and more for voting and participating in the daily challenges. Midjourney IS the best, but you will pay for every credit there.
some key prompts to improve your work without having to reference Artstation( using styles always does) are
Beautiful Composition, Back-lit, Realistic Style, Hyper Detailed
enjoy yourself
This can be scary for celebrities! They were actually married, but what if they weren't! I input Angelina Jolie and Brad Pitt kiss at an awards event. I have no idea if they actually did IRL, but they did now. Celebrities and politicians are screwed, Actually anyone can upload two people's photos and make them kiss or worse!
I have already seen far too many passionate AI embraces with tongue between Donald Trump and Joe Biden to traumatise me for life on social media
Now I tried Brad Pitt kissing Paris Hilton. Not perfect bit it probably will be by V6! I don't mind AI being used for art but I do mind it being used to trick people in REALITY!
All this will do is make sure no photo will ever be trusted again. That's actually probably a good thing for politicians. They can just claim it was AI by default.
There are already several flaws in those AI kissing images I can point out. Hopefully it's never perfected.
This is version 5 which is way more realistic than version 4. Eventually there will be vs 6. 7. 8...
music clips are going to be crazy
Oh boy, I've really fallen into this rabbit hole we call AI art. Recently I've been playing with Stable Diffusion, and I'm so much loving it. I recently bought a new computer, so I'm finally able to run it locally. I've always wanted to to be able to turn my Daz renders to paintings and comic style, and since I'm totally hopeless with my drawing skills, it's AI to the rescue.
At first I had tons of problems when I tried the img2img and that ended up with a disaster, but as soon as i found control net ( for me it's canny and depth models where I started to make progress ), where AI could first make an outline from my Daz render, and then turn it into cartoon style art. Currently I'm trying to learn how to use Lora, since I need consistent characters. If I'm not totally mistaken, with Lora we are able to teach AI about characters, outfits, locations etc. and when the ground work is done, AI should be able to make entire stories using those consistent characters in familiar surroundings. I'm sure somebody who knows more about this stuff will correct me if I'm wrong, but that sounds like a dream come true if I'll ever get that working.
The one major thing I've noticed is how fast this techonology is evolving. Lawsuit or not, I'm afraid it's too late to stop this train anymore. At least stable diffusion is open source, and it's already out there, and I seriously doubt people will stop using it no matter what the court says. That gets me to the next thing, which is the amount of people who are investing their time and effort for this is just mind boggling. It's scary really. I was thinking month ago how cool it would be to pose your own characters in stable diffusion, and apparently now you can. Well, not DS level posing, but basic stuff, and it's only getting better every day. With so many people working with this, I don't think it takes that long before we can do so much more with it. If I was Daz CEO, I would jump on this train right now, and started making some kind of addon in DS where AI could read 3D models straight from the studio, and then "render" stuff.... well, I suppose we all know by now how fast Daz is making new stuff, so that's not going to happen, but guy can dream, right?
Oh, and I forgot to add some pics. Here's from Daz to comic
1. Lawsuit is against abuse, not against "using ai".
2. Maybe things can go through the roof, but it's not magic. First and foremost the biggest improvements likely will come from combining different tools, rather than "fixing this one". The point being, that this isn't a classic computer program, where you could "fix the bug" (... in theory. In a way, machine learning is the answer to not being able to "implement the feature" with classic means, disregarding some other statistic technique for the moment). Posing may be possible, if you have enough training data tagged with posing information. Ironically, who does the job? (And who takes yours?)
3. The magic image generation from a prompt is different from coherent storytelling by nature, it would likely have to be a combination of tools, or say, a combination of more specialized tools will give faster and better results than "prompt engineering" (at some point).
4. Combining (mostly new, but also "old") tools might have endless possibilities.
I've added MJ to a private server on Discord, but the results are practically awful. The faces are... Well, not faces.
Are MJ generations worse on/in a private server?
depends on what artstyle you want to follow, if anime are your thing there is a ton of civitai models around there, the bad thing is more difficult emulate a face in anime style, in my case is photorealism and semi realism, I do img2img a lot, instead elaborate a complex prompt of 525 words I do this line in img2img:
beautiful studio soft light, rim light, vibrant details, luxurious, lace, hyperrealistic, anatomical, facial muscles, elegant, beautiful background, octane render, 8k, best quality, masterpiece, illustration, an extremely delicate and beautiful, extremely detailed ,CG ,unity ,wallpaper, (realistic, photo-realistic:1.37),Amazing, finely detail, masterpiece,best quality,official art, extremely detailed CG unity 8k wallpaper, absurdres, incredibly absurdres,
no need to explain what are you want to create as "female general commander with red beret and black jacket, asian etnicity" no, because the render is selfexplanatory, you just declare quality tasks in the prompt,
then you insert your image or daz render and voila:
the same with images, just define quality parameters, no need to explain what is on the image, all uses Karras variants but I use DDIM, it does wonderful color mixes:
for celebrities, there is Loras available, I use kpop celebrities for example+img2img+ddim+basilmix
If AI can solve the problems with unrealistic joint bendings, expressions and hairs on 3d characters, I'm all in.
If not, I won't care about AI.
I am trying PIFuHD now but not having a lot of success
one image has generated a recognisable 3D figure so far out of a couple hundred I tried
yes it is an AI image too, my photos came out awful
Zremeshed, polypainted with G9M
video
this should solve 99.99% of the ethical issues anyone has with AI
I am not fond of subscription services but would certainly toss money at this over subscribing to another nonethically trained AI service such as Midjourney
of course using Stable Diffusion on my own computer for free is the cheapest option
and for my own personal use I have no qualms whatsoever
but for commercial use Adobe is offering the answer needed and others are likely to follow suit
This actually looks more like the tools i'd envision. More control, a combination of tools (not all magic), and so on. How diverse the results will be, remains to be seen. Also how good it will be in the end. Of course i could imagine the fully voice controlled, pretty language-savvy combination of such and more tools, so you stand in your living room, or someone's, and you work babbling to the wall, that the results get projected onto...
That being said, rest assured, that it doesn't solve as much of ethical issues with "ai" (in general!).
1. I assume so, but i don't know, if it's ethically trained. I don't recall any passage relating to such from that video, other than "commercial use" or something. Is copyright even mentioned?
2. You already mentioned "commercial cloud service" - this may be fair, but it's certainly more than a 0.01% issue ;). E.g. concentration of power and control are generic problems, which always concern ethics.
3. In a broader sense: if concentrated services remove jobs at large, we will have question of societal scale, some of which are just stability and survival in the first place, e.g. if they remove abilities from societies at large, some will be "just ethical". Typically that's the joke, because questions are "ethical", until deadly at large.
4. So ethics also concern how we let stuff happen and form or deform society. The most immanent part is abuse with training data and the output of the system (flooding with lookalikes, direct competition with artists using their style, stalking, manipulation...). The longer term societal effects can't be fully foreseen with image generation or hybrid tools. Likely there isn't much you could do about it anyway, other than inventing new taxes for large cloud services of specific kinds. Of course you could regulate training data, copyright, some conditons for commercial use, and some of the output, e.g. concerning lookalikes and styles of living artists, if "stalking" with lookalikes becomes a broader issue.
And just to be sure: it's not "the ai" (in general), though what you mean, probably is that there is something cool that doesn't seem to be based on potential forms of abuse. For an educated guess, it's just a system of systems, with multiple differently designed "ai" subsystems, likely also with some more classic stuff inside (not just neural networks), regardless of what other systems consist of. (Human intelligence may be such/similar too, at least in a logical view, but that's something completely different.)
So elaborating on 1., checking the adobe page: "our Firefly family of creative generative AI models is trained on Adobe Stock images, openly licensed content, and public domain content where copyright has expired. Training on curated, diverse datasets inherently gives your model a competitive edge when it comes to producing commercially safe and ethical results."
a) Stock images - looks good. Is it Opt-in? I hope so. (Time limit for research reached.)
b) Openly licensed images - foggy day! What does that even mean?
c) Copyright expired - ok, that's... ok, i think?
d) Somehow diverse data sets - not sure that's an addition to a)-c), probably it's just a space filler, probably generated by a language model.
(Concerning 4., there is a fair lot of text about ethics of the output, which may or may not relate to contemporary artists. The interface i saw in the video looks as if they could in theory try for or have designed towards mitigation/control of such issues, but it's not certain, as the prompts aren't really a topic. Further research needed...)
(At least that sounds better than "we take everything and do what we want anyway, except for depicting cannibalism and nudity".)
oh I know many will not be happy just like they were not when the industrial revolution took all the jobs
and I sympathise
this however addresses the legal issues
doesn't mean a Skynet future is any less uncertain
I never said it was a good thing
there are no good things on the horizon
the best scenario for life on Earth if not human survival is a reset like the meteor that took out the dinosaurs or a pandemic, massive electromagnetic pulsing sun flare or something catastrophic that won't destroy everything