Writing on the wall for digital content?

#1

Saw this today and gave me a bit of pause.

Feels like an artists world is getting smaller and we’ll all just be making content for the world, let alone creation software like Cinema.

#2

This looks like a load of horsecrap to me…I think it would be an understatement to say that they’re exaggerating what this system is capable of. We are apparently supposed to believe that these scenes were created by people saying things like “make the lamp much larger”?

What lamp? Who made it? Are we supposed to believe all the unique, detailed models in these scenes were spit out by an AI while someone spoke?

And who lit these scenes? I smell a rat. As someone who’s been in the game industry for 28 years, things have only gotten more complex, not easier! Where’s my “make good art” button?

I think the biggest tell is that we’re not ever seeing what’s happening in the VR space while they’re “directing”.

#4

It’s always far fetched until one day when it’s available for everyone.

I remember when matchmoving took ages to do, now a phone will do it in realtime while lighting and shading 3d inserts.

We can all remember the days before content aware fill.

The thought of speaking to a computer and getting weather reports in the living room was Jetsons like a few short years ago.

Look at the amount of money being spent on AI, billions and billions. You’d have to have your head in the sand not to believe we’re on the verge of a new industrial revolution which may see many of us side lined.

There will be shysters popping up to take the money of the ill-informed or ignorant along the way but way to not eventually be out of a job is to stay informed and hopefully ahead of the AI game. The AI genie is out of the bottle so we’re going to have to live with it best we can…until it self replicates killer robots and wipes useless slow flesh based lifeforms out of the way.

#5

Interesting new use of UE4
https://www.quantumcapture.com/ctrl-human

#6

Photoshop can fill in blanks that you lasoo
Google has very reliable voice recognition
Deepfakes can insert people into videos
Nvidia’s system can draw landscapes from 4-5 coloured lines for hills, rivers etc

Throw in some premade assets and there’s no reason this couldn’t be done.

#7

I think you guys are hallucinating. Look, I get it…we’re absolutely going to see major advances in AI, and many aspects are going to become easier because of it. And I don’t have a problem with that. I do have a problem with the smoke and mirrors of this video, insinuating that by talking you can create scenes like this. At best, we’re talking about a VR content browser where you drag objects into a room.

At the end of this video, we see each person say “I made this.” Really? Maybe they should say “I arranged the furniture.”

Remember the gist of this is that they’re not just creating environments, they’re creating game environments. Who’s handling the collision detection? How about the pathfinding? Who optimized it to run at 60fps?

There’s a whole lot more involved here than content aware fill and google voice. Maybe I’d be more convinced if they showed what’s going on in that VR stage, but they show nothing other than final renders. With pre-existing assets.

#8

Quite true, and I’m sure you’re right which is the major reason they are not showing what’s actually happening in the headset, but what’s bothering me is people like this are getting millions in investing to take things in this direction and further. But within 5 years we’ll be heading toward artists are just working to fill asset bins for things like this instead of artisan projects from scratch. Just doesn’t feel good to me. I’d like to have that for our own pipeline in what we create but it feels like a tool to cut out creatives altogether once “content packs” are available. Mash has excellent points on all of the latest AI fads as well. In one respect, it is a great time saver for artists, but in the other, it’s cutting creatives out. This will head into coding as well. If they can perfect speaking to making the objects have collision detection and direct interactivity… If you don’t think that’s coming, you’re fooling yourself.

#9

And Prometheus wants a product recall…he meant to give us fur…but the AI misheard him.

#10

This community sinks further into the mire each day…

I think I’m at the point where I browse it more for entertainment than information. Pretty sad state of affairs given how respectable it used to be.

#11

While i think that AI will have a profound impact on digital content creation, we are still a long time away from high level functions. The Promethean video leaves me with the impression of an investment pitch, not actual technology.

#12

As they say: The immediate impact of technology is almost always overrated. The long-term impact of technology is almost always underrated.

#13

It definitely comes across as that, but the intent is there. Also, define “a long time away”. In other news, along these same lines, Magic Leap is looking for a buyer…

#14

This is was it does. Nothing special. Nothing new.

#15

Deep Fake and similar uses have been around for over two years and while they became more refined they are still a good step away from beeing usefull for everyday work. The most used stuff is what Adobe integrated, content aware fill etc.
These have all become a lot better, but it all is a far cry from what we can glimps at if we watch https://www.youtube.com/user/keeroyz
For high level use like the Promethean pitch i would expect no less than another 2-3 years before single tools become productive. A mostly AI based workflow is still further away and might not become available in my active worklife. The number of specialty solutions will soar though. Just my €0.02 and i might be completely wrong :slight_smile:

#16

Srek how hard is it for a company like MAXON to integrate features like those in that youtube channel ? As far as I understand scientific papers provide all the technical recipes to reproduce their outcome. So why do we wait so long till their technology become available ? Do those scientists ask for copy-right privileges ? I understand when Disney or Pixar publicise papers. Yes, they retain their right to have the monopoly of their technology but is it the same with university scientists ? I think the Disney case is on the “patent” spectrum but the others are on the “free science for all” corner …

#17

Most papers show stuff that is not protected, often they come from academia. In the US software patents are possible, but most papers are not covered by them, on the contrary. The most current papers are not related to patented algorithms since patents take time and talking about stuff that is not yet patented easily invalidates your patent. So if you want to make a name for yourself you publish instead of apply for patent. Think of papers as ads, not cashcows.

Papers are often very specific and only solve a very narrowly defined problem. While for that narrow application the solution is impressive, it can fall apart under even slightly different circumstances. It is often the case that parts of papers make it into products rather quickly and quietly as under the hood improvements of existing functionality.

#18

Lol, I knew it. What a ruse this video is. Or as Srek put it more nicely, an “investment pitch”.

Besides, we already have this. Show Me Blue Chairs

AI Magic!