AI is going to be our biggest game changer
ghastlycomic
Posts: 2,531
in The Commons
More than better, faster render engines I think AI will be the biggest game changer as far as 3D computer animation is concerned. I have a feeling this will mean instead of being modellers and animators we'll all become directors. Can't wait.
https://www.youtube.com/watch?v=ZXFmZsv0Ddw
Imagine when we get to the point where we're able to describe to our computers what we want and then work on refining how the computer interprets our wishes.
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
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
It's nice to know that the ability to make crap more realistic will soon be at our fingertips. It's about time Autotune came to the visual art creating masses.
What's AI?
Good for some things maybe but not a fan.
Artificial Intelligence, and honestly I can't see much benfit for it in DS
I understand some people use artbreeder to generate multiple 2D images from a single image.
Writers will become the most important collaborator in the process.
Yea, I don't either, but then no one's ever accused me of being much of a visionary ;).
Laurie
I tried the animated GauGAN too but cheated and used a second AI program EbSynth to generate the inbetween frames.
it is the opposite
I took 3D footage and art styles to generate an art animation
well to be fair we already have the google deep mind learning how to "anime characters 3d alone:
also others works with AI and animation, it's really a matter of time like 20 to 30 years more and games, movies and renders will be made by just you "talking and the robots or AIs doing all the job while you just watching,
There is now a lot of content made with professional real live actors that isn't worth watching. The biggest challenge is being able to craft and tell an interesting story that connects with audiences, and bring it to life on screen. AI isn't going to change that. 90 percent of everything is still going to be crap.
At the moment I don't think that AI is capable of making up for natural stupidity in my case. It would be nice if it was..
I'm really not sure where it's going to go. The only thing you can be certain of is that the computing power needed will be huge, maybe too much for it to be home based, and you may be telling an Alexa or Siri or whatever, and the computing is done remotely through a subscription service. If so, not for me, thanks.
Regards,
Richard.
Can't wait for the neural interfaces, no more keyboards, mice or monitors, you just think and the computer creates the whatever you thought straight into your brain...
Nvidia built Tensor cores and included them in even their consumer GPUs for a reason. The first thing it can do is open the door for better denoising and upscaling. Most denoising is still pretty "dumb". When DLSS is properly applied to video games, they can use Tensor cores to run a game at a much lower resolution, thus a much faster frame rate, while Tensor rebuilds the picture in a larger resolution. The first iteration was not great, but DLSS 2.0 has impressed pretty much everybody, even the most hardcore sceptics. With good DLSS, they can actually create a picture that is clearer than if it was rendered at the original native resolution...that is just crazy. It is really exciting.
Now if that form of upscaling can be applied to Daz Iray...just think of what can be done with that. You could render an image at 2K and upscale it to 4K and get all the details it would have if rendered at 4K. And obviously that would render much faster.
Perhaps in the future AI could be used to better animate your scenes, so that you do not have to do all the little stuff. Like a walk animation is not just purely canned, it is fully adaptive and can adjust itself to walk over different surfaces, with the model correctly balancing themselves. Some animation can do this, but they still rely on preprograming such features. What I am talking about is the software can do this own its own without that. You could create an entire crowd of people who all automatically follow a routine, all the characters are also unique but dressed for the environment...and you didn't have to do anything other than tell the software how many people you wanted and what their general looks were. Then it goes and creates all of it for you. You already have AI creating faces, so this would be the logical next step. Then you can focus on your main cast and story.
You could have an AI create an entire city scape for you, with everything in it to your tastes. Want a cyber city, just tell it what you want. It goes into the database and creates a wholly unique city for you. Then you think to yourself, hmm, I want more neon. Ok, you got it. You can just dial it in. It can make use of all of the resources you have, so brick textures are used for brick, and so on.
In the '70s after I turned down a job offer from a friend in college to help him make video games, I confidently predicted "there's no future in computer games".
I bet the female figures will still be cross-eyed and carry the permanent 1,000-yard stare... simultaneously.
Another set of tools for the toolbox, but I don't want to give up control. Hoping that it will make animation easier, faster, and more realistic.
As several folks mentioned above, I am anticipating an AI package that would take a novel or screenplay, and do the heavy lifting/tedious parts to set up the basics to make it an animated movie, but you would still have to make the characters, etc. look and move like you want. Even if only used for storyboarding, it would still be useful. I would want to be able to have the option to do more or less detailing and character design, depending on the project. I think it would be very interesting as a writer's tool to help visualize scenes during the writing process.
As an example of existing technology, the program MASSIVE has been around for about 15 years, and has been used to help create large battle/crowd scenes in a number of films, Like the Lord of the RIngs series, the Narnia films, the Marvel Avenger movies and more. The results are pretty impressive, I think. A combination of MASSIVE and VUE INFINITE was used to semi-automatically generate huge ecosystems for Cameron's AVATAR. They have made scenes with 10,000 individually animated "agents" with automatic variations so that no two are alike!
http://www.massivesoftware.com/index.html
http://www.massivesoftware.com/gallery.html
Demo: https://www.youtube.com/watch?v=cr5Cwz-5Wsw
A single one-year license for Massive Prime is $16,000, or you can rent for $2500 per month. Way too rich for my blood, but in a CGI studio setting, it can be a huge cost/time saver!
I will wait for the free Blender Plug-In
And another current use:
https://www.space.com/moon-landing-footage-remastered.html?fbclid=IwAR03TXIh6IDZ9lu8zIu08r1Cw-P3h4PkJptbssSJBRUsKlBAVDu92SB1e4I
The AI restoration of classic films and pictures will only get better as the technology evolves.
While I do believe AI could help create art, I do not believe AI itself could in the forseeable future actually create art. It would still need hands on direction from a human being to accomplish that. And, the more control the creator of art leaves for the AI to decide, the harder it will become to actually create art. It's just the nature of art, to go where noone has gone before, to both define and break conventions within the same piece. AI may have the benefit of ground rules within its programmed frame of conciousness, and randomizers to break those same rules, but it won't be able to recognize when it is "onto something". It can spew out a thousand different pieces, but never recognize the brilliance of piece number 738, and just offer it as a one thousanth piece of a collection of otherwise soulless products.
Almost Intuitive.
Actually Impossible
All current AI algorithms are nothing more than statistics ones and all of them work at reducing a given error measurement or multiple errors measurements. And that's it. There is no real "artificial intelligence". Under the hood it's just some classic math and nothing really new. Quantum computers could do the magic because how they work really is quite magic and although they work they really shouldn't but they do.. so real AI might come from those in the next 50 years. But for now AI algorithms are just math as is explained on this link for example: https://dev.to/liashchynskyi/how-neural-network-works-lets-figure-it-out-32o0
Amazingly Interesting
I'm one of those actually depending on DLSS to upscale and on the new AI to help assist me in making decent animations. It should make things quite fun. I will be buying a nVidia GeForce RTX 3070 8GB in January (if any GPUs are available then) after I save (yet another) $600 to give the new HW & SW a go as up until this nVidia 30X0 series release what I've been needing hasn't been possible let alone even available. However, $499.99 is my limit on what I will pay for a GeForce RTX 3070 8GB as that is the manufacturer's MSRP and I refuse to pay more. With only 8GB I may need to composite (and may sensibly should composite any way to really speed up render times) but I'm hoping the DLSS upscaling will let me render in 8GB RAM at 720P or 1080P and then upscale to 4K. I guess I'll find out but that's the whole point of DLSS mostly. Ugh, I've dropped so much money into this but at least the HW and SW has finally reached a mature capability level so that it now makes economic sense for me to have spent that much money of all those things because I definately can't afford upgrades every year!
I sort of want to get a nVidia GeForce RTX 3080 10GB for an extra $200 but only getting 2GB more makes that unlikely; those extra compute cores are tempting though and put performance firmly better than the 2080TI while the 3070 is only slightly better than the 2080TI. I'd like it if the AMD Big Navi 16GB (6900XT) for about $500 had a SW stack as capable as nVidia's but they don't and it could take years, if ever, for them to develop those software capabilities to the extent nVidia has done.
Art Illustration
The way this works is that AI will ease the creation process while also raising the standards. 10 years ago a professional 3D artist would be considered at best a mediocre one now.
It all evens out. Toy Story 1 was revolutionary for its time but now it's subpar. Both Toy Story 1 and 4 took the same amount of time to create with the latter costing significantly more.
Wasn't that what we heard 5 years ago? And 10 ago? And already 20 years ago? Or already in the 1980's when Cyberpunk was the THING.
My problem with AI is, that it's humans that try to achieve it and build it. And as humanity as her own problems with intelligence... well... nah, don't see real AI soon (tm)
A reversed donky sound.
AI is going to big thing in future but if real AI ever made then it will definitely compete with human as AI can improve themselves. The day will be milestone for AI when it start writing poems.
But Artificial Ignorance runs the world.