AI is going to be our biggest game changer
This discussion has been closed.
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Oh, a whole lotta AI on this one.
Music video. I think 4 or 5 different apps.
Here we go.
Google DeepMind Scientists Claim that AI Could Destroy Humanity; Are You Buying It? | Tech Times
Thanks for posting - so it looks like, the problem will be solved - sooner or later.
That is interesting - made by the professionals: The Museum of Modern Art
AI Art: How artists are using and confronting machine learning | HOW TO SEE LIKE A MACHINE
video
I stole all the lollies
even they are only 50cents
What I would like to see is AI within DAZ Studio, so I could load up and dress my characters and props, then use English to tell it what I want with regard to posing, animation, and camera travel. And have it figure out how to light the scene appropriately as the camera moves through it. I guess what I'm thinking about takes a scene and creates animation tracks for it, so that you can save them separately and tweak them if necessary until you've got everything looking just like you want with a LOT of help from the AI tools.
Buying content and dressing up the characters is easy. Everything after that takes a lot of hard-earned skill for a human, but might be easy for AI to figure out. And the stuff AI messes up, like the number of fingers on a hand, would not be a problem if it's working with existing 3D content instead of trying to create the characters as part of the image.
I'd love that too but if it happened, Daz Studio AI would probably not run on my video card which has only 6 gigs of vram. And since it's a new computer I've no plans, moneywise, to upgrade. Be different if I was using it for work; for a hobby, no matter how enjoyable, I simply cannot justify the expense. :sigh: Maybe someday, if I win the lottery ...
I did a render of Ghengis Khan in DAZ Studio... used an AI generated landscape as background, and painted the two together...
Nice work!
Well today it's built right in the Daz! LOL
I have applied Make Art button from yesterady on https://www.daz3d.com/chihuahua-for-daz-dog-8-penelope-and-zorro-add-on
and get this below. Works very well for me...
... and that one, too ...
They could call it "Skynet: Matrix Edition".
You're basically looking for the Holodeck computer from Star Trek TNG, so that you can tell it which Dixon Hill story you want, or how you want London to look in your Sherlock Holmes adventure. Just don't deactivate the safety protocol!
For the comparison - the artist watch the reference image on the computer screen and then paint.
How a Professional Artist ACTUALLY makes Paintings
How to Project Photos for Painting
Do a search for LED Light Pad - very useful for turning renders into something to paint !!!
Lucy
Professional artists projecting photos for painting is modern mechanization of an ancient practice of camera obscura or lightpads (or literal windows, like, windows in buildings). All those painters are also capable of drawing from memory. It's not even remotely comparable to automated image generation.
But take as much as digital brush stroke stabilizers and working layers out of digital art and things start hitting the roadblock.
So maybe one could see AI images generation as:
"modern mechanization of an ancient practice of camera obscura or lightpads"
- AI postprocessing with the image provided
"painters capable of drawing from memory"
- AI generating image from the text prompt
OK, now I'm pleased again... all it took was to determine what AI helps me with and what I still want 3D renders to do.
Scenario: Minerva (the Minerva 9 character in the DAZ Store) has been sent on an errand by her lord in the nearby castle.
The Problem: Minerva just didn't look like Minerva when I let the AI try to adapt to suit the changes to the rest of the scene. (Incidentally, I had deliberately built a castle that was not the aesthetic I was looking for, as I specifically wanted the AI to riff off the composition I created in DAZ Studio but wanted something more unique looking than straight-ahead model renders.)
The solution: Do Minerva as a separate render and save as a PNG to preserve the alpha data, then use a separate render of the background landscape to feed into the AI to work with.
Then after landscaping with the AI and blending in Affinity Photo, the finished piece looks like this...
(What I love about this method is the level of instant control over the composition - no unwanted duplicates e.g. castles on every single hilltop - and could then focus on what style I wanted... saved me so much time.)
Looks nice, @Jabba.
I have also tested AI postprocessing of renders and I like the results, as well.
The key is to find effective and not so much time consuming technique to blend images.
A peninsula memory painter at your service, having "learned" from all painters of all time. Or a shelf lower: a nifty cover-painter with an encyclopedia at hand. You tell them what to do, and they perform according to their liking and their corporation's guidelines :). Further they keep a copy of all you said, the image, and their corporation ensures to track it all over the internet ;). Well, anyway...
Not sure what you are trying to say.
That people using AI-generated images are on same level with artists capable of painting entirely from memory because said artists sometimes use aids to speed up their work?
They aren't. Take the AI away and they have nothing left.
In comparison a professional artists might use aids and references, but if they can't paint without it they wouldn't make it through the art school. They still need skills to pain following the projection with physical paints and brushes.
The AI itself also obviously is not on the same level with artists since it's a random image generator. Just that you roll the dice enough times to get what you want from the prompt.
I know some people start being emotionally invested trying to claim they can create because they spent so much time using the image generators they want/need to believe. But let's not fool ourselves. It's not the same level.
I am a lousy artist but I can plonk a few premade 3D assets in a scene and hit render
(I am probably not great at that either
)
I am not claiming anything other than using AI is easier and looks heaps better
for those just wanting pretty pictures of subjects of their choice it's the bees knees
which is the vast number of ordinary untalented people
disclaimer, not actually defending AI here just saying why it's not going to go away
That image looks great, and I really like your workflow there.
Here's more or less how I see it (and it's roughly in line how US copyright office seems to perceive it).
If someone takes an AI output and edits the most of it, that's probably creative. There's going to be lots of edits to get exactly what you want. Probably the entire image. From shapes to colors to where it ends on the canvas.
And that's okay because at this point this is comparable (in terms of amount of work) to doing collages or photomanip art. It doesn't matter much if the source assets came from 3d assets store, photos, older art or whatever.
But taking a prompt output, then maybe cleaning it from errors it's less an artist work, and more an editor work. The Ai is the creator here, the human works editorial on this. Majority of the AI art everywhere seems to be exactly this. The least resistance route.
I need AI to make my hair atachments look like actual hair.
with inpainting it definitely is possible
For a direction with computer games, you could almost go entirely meta, by making use of language models and image generation, as well as language model processing given images as well. Flatrate and some interfacing needed, though.
- MORPG with believable wording, or correction of player language.
- MORPG: Paranoia - finally feasible?
- (Telling stories from facts collected from the in-game history of a character or autonomous agent.)
- ("Cinematic" renders, WYSIWYG +- you bet.)
You could in theory throw people into a random story that evolves almost entirely based on human-machine interaction. Sometimes could give players choices or allow a question, somewhat guarded by context.