AI is going to be our biggest game changer
This discussion has been closed.
Adding to Cart…
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.You currently have no notifications.
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
Sort of like the Block Chain, the company that wrote that software probably has an incentive to slap the AI moniker on every product they make. I'm sure it does drive sales.
But before Neural Nets, many things were called legitimately called AI, because while they were still algorithmic, but it could take that algorithm that is really very simple, but follow it with perfect precision for millions and millions of iterations extremely quickly, something humans obviously cannot do. So they can solve problems that a human could not because of purely practical considerations. That is, you don't need AI to solve a 2 x 2 Sudoku puzzle. But given enough time, an AI that is the same algorithm could solve a 2000 x 2000 Sudoku puzzle and it would be provable that the AI is eventually going to solve it.
Before Neural Nets the canonical first example in an Introduction to Artificial Intelligence course was likely to be the recursive algorithm that could solve The Tower of Hanoi problem or the depth-first search algorithm to solve that game where you have to slide the numbered tiles around to put them in ascending order when there's only one missing tile in the grid for you to maneuver with; I can't remember what that's called or if it even has a name.
So AI did not start with Neural Nets, and they did manage some pretty impressive feats that sometimes made them seem superhuman or magical.
Let us not get into quasi-religious debates about AI in its various meanings, please.
There are so many uses and potential uses of Artificial Intelligence for 2D and 3D image and video processing and art generation on Daz Studio content...
The combination of deep learning/machine learning, computer vision, object recognition, neural networks, GANs (generative adversarial networks), AI denoising, AI upscaling, AI image classification, audio-driven facial animation, NLP: natural language processing, speech AI, Transformers replacing CNNs and RNNs - Transformer Models that can translate text and speach in near real-time, image inpainting, real-time motion capture from video applied to 3d character rig, pose estimation, computational geometry, photogrammetry processing of just a few photos into a 3d model using neural rendering... Neural Radiance Fields: NeRF... using AI to generate 3D models — Getting started with NVIDIA Instant NeRF, NVIDIA Research AI demos: https://www.nvidia.com/en-us/research/ai-demos/
Control variates in NVIDIA Ray Tracing AI... Geometric Deep Learning looks really interesting.
In the near future, you'll be art directing A.I. with text prompts (or just speaking to it in a natural language) and describing the scene that you want it to generate.
Maybe one day we'll just talk to DAZ Studio, tell it all the content we want assembled in the scene, describing the rendering and lighting to be used, the colors and textures, the context for all the poses and camera angles. Maybe with some intelligence, our content libraries will be self-organizing systems. Maybe our renders will feed back into the system and maintain a consistent look of our work.
I think AI is very similar to evolutionary instincts. Instincts are just pre-programmed behaviour based off of statistical probabilities of past events. So I think calling AI, AI, is very relevant.
You can do that with the models today. I doubt Daz would be enabling this type of AI powered workflow in Daz Studio without them changing their business model.
Exactly... as soon as arguments reach the "prove it doesn't exist" point it's time for me to leave the building...
You inspired me, Wendy. If you need something quick and dirty for something that the viewer is not really supposed to look at too closely, Midjourney is viable. Sorta, kinda.
https://youtu.be/ZAR4PTmMEMc
I am rendering backgrounds mostly in it too
you all can have this
Ooh, that's a nice one, Wendy. We should have a contest where the objective is to guess what the text prompt was :)
When I have a minute, I want to make a mask for the stalagmite in the middle, and there's a plugin for Blender that will infer camera characteristics from a photo. All this because I so want to walk around in that image. Thank you for my new obssession, as if I needed a new obssession.
stalactite Stalagmite cave vr360 --w 2000 --h 1000
That was a tremendous help, Wendy, thank you! Where did you find out about all the command line options?
What does specifying vr360 do? I'm not sure I appreciated any difference.
read discussions, mostly https://www.facebook.com/groups/midjourneyai
looked at other's prompts
and not what I hoped (Spherical lens) but seems closer to than without
Thank you Wendy, I joined. Some really good stuff in there.
I have reached my limit
also
I hope they leave Discord and get a Dedicated Secure Website with Paypal as an Option
my bank balance doesn't share my views, my cards have all barricaded themselves in my wallet, I still manage to scrounge coins daily for DAZ though
Agreed; a dedicated webapp would be nice, with a better interface for doing this type of thing. Working with multiple version of images just does not fit the metaphor of a top-to-bottom conversation.
Have you tried that advertised "/relax" which reputedly gives you extra goes? Or doesn't it work?
only on $30mth or more
I have added $5 of metered top up now as some credit left on my Visa giftcard
...and
I almost can do spherical renders in Midjourney
prompt garden Ricoh Theta-S equirectangular vr360 --h 1000 --w 2000
https://www.facebook.com/photo/?fbid=10225986887972796&set=pcb.10225986889052823
https://www.facebook.com/photo/?fbid=10225986887972796&set=pcb.10225986889052823
Oh wow I must remember that prompt. I've been totally hooked on Midjourney and would love to get a HDRi out of MJ.
Thank you for sharing :)
Anita
yours look great
as long as one obscures the seam with something thet should work, Gimp can save as HDR
this may address some concerns about copyright and usage
disclaimer: I am not a lawyer
https://www.theregister.com/2022/08/01/david_holz_midjourney/?fbclid=IwAR2uSxJ0BCcfn9iUTdI4p7rRa0MUFJ1VpEA2Tz3F6Mieyw7Ap_BIVkAuV8Y
By far Midjourney spits out some of the most interersting to date and another potential is Disco Diffusion which takes some DIY to set up.
Staying on topic. If used as just another tool it's certainly very possible.
Using Midjourney for concepts to create ART in Blender -
[MECHDUSA] A.I. Art + Photoshop | Photobash Timelapse -
Using Midjourney to create concept art for 3D models -
accepted into Dall-E 2
make art button indeed
Somehow these pictures remind me of Cecilia Giménez's attempt to restore the "Ecce Homo" picture in the Sanctuary of Mercy church in Borja, Spain.
When looking at them, I'd ask myself: Did the artist intend to look like this?
There is that other program from Nvidia called Nvidia Canvas, that does some interesting things for 2d images..
Most of the stuff shown in this thread only provides me one syllable for description. That's "Ew." If this is the future of art, you can have it. It's garbage.
The future isn't a single distinct destination when at some point we can say "We're here, we're in the future...", and so the point it not about what the current state of the art is, but its trajectory. I've noticed that the people who participate in or follow technology understand this better and are more excited about it. When you've seen the faces generated by MJ v2 and you compare them to MJ v3 today, you really have to wonder what v4 is going to be like. We'll have to revisit this in six months, and then again in another six months, ad infinitum.
I am also reminded of that funny line from the movie JFK when Jim Garrison says that his dog is not very smart because he could still win two games out of three against the dog at chess. Some are completely missing the point that a computer synthesized an image from a text prompt in a matter of seconds, and instead focus on its imperfections as if the pace of improvement were not as incredible as the technology itself.
Am I now supposed to congratulate the program, that managed to do this task, or maybe rather the programmer(s) who managed to produce it's sourcecode?
But of course you are right. People who work with computers for, like, 40 years or so, probably have a totally perspective about these "leaps in computer evolution". From my other hobby, strategical combat simulations, I can declare, that in that area nothing that would deserve the distinction "AI" exists. Artificial, yes, but intelligent not so much. Most strategy games still have to fall back to having the computer controlled player cheat to offer a somewhat interesting gameplay. The more details are involved in a game, the less intelligent the "AI" acts.
So between trying to emulate a human being in a strategic game and searching the internet and putting some pictures together to fit a short description by a human user... still a looooong way to go for the "intelligence" part.
I can really only show some of mine and apparently nobody can reach the Facebook group I linked with the cool suffs by others
my subscription has expired
but here are some, mine are quite lame compared to others
forum lost half my uploads
try again