http://www.youtube.com/watch?v=00gAbgBu8R4

Well, thats one small step for the gaming industry, one giant leap for the Matrix

8/2/2011 12:16:11 AM

that video was actually compelling. I have to wonder what kind of recursive math they're doing to render like that. mathematically i can't think of an efficient way to do that

8/2/2011 12:49:29 AM

Vaporware

8/2/2011 12:51:22 AM

Let me know when they're handling object interactions

8/2/2011 1:19:07 AM

I deem this the official tech demo thread.

8/2/2011 2:03:07 AM

Quote :

" I have to wonder what kind of recursive math they're doing to render like that. mathematically i can't think of an efficient way to do that "

My guess is something akin to jpeg encoding.

3d equivalent of fractal patterns baked into the engine, then procedurally generating objects in real-time based on their encoding in this scheme.

Procedural generation has been the buzz word for a while now (Sony had been pushing it to overcome poor memory bandwidth between GPU/RAM on the PS3 IIRC), but AFAIK no games are using it in a fundamental way.

It seems like the next logical progression would be a fully procedural engine, since modern GPUs are getting more programmable and have more complex computation hardware.

8/2/2011 2:31:08 AM

My bad. I clicked on the thread expecting it to be about how industry is shifting to micro transactions that take place after the initial game purchase, aspects of a fully developed game being withheld to drive DLC, and ever increasing DRM schemes that have the goal of destroying the used game market. In the next 10 years I wouldn't be surprised if all games require a net connection so they can charge you by the hour.

But this was a new tech thread. I saw that video somewhere on tww before but I can't find the thread. That was months ago, I wonder if they're any closer to getting this tech into an actual game?

[Edited on August 2, 2011 at 8:31 AM. Reason : -]

8/2/2011 8:30:42 AM

Tell me that accent is fake.

Also, if I have to hear how they're not artists and not a game company one more time....would it have killed them to hire an artist? If their engine is as the shit as they say it is surely they could find some investors.

[Edited on August 2, 2011 at 8:51 AM. Reason : .]

8/2/2011 8:47:20 AM

^An Australian with a terrible head cold... or a crayon shoved up his nose.

8/2/2011 9:15:08 AM

^I was thinking a British "surfer dude" with homosexual tendencies

8/2/2011 9:19:47 AM

Quote :

"would it have killed them to hire an artist"

I think they do have artists, just not very many. He certainly didn't say they were all programmers.

8/2/2011 9:25:22 AM

skeptical of this be I

8/2/2011 10:08:13 AM

VaporNonsenseBullshitware

8/2/2011 10:44:24 AM

Vaporware

8/2/2011 11:54:30 AM

this will go the same way as intel's ray of light processing... or whatever they called it. Too lazy to even google it.

8/2/2011 12:13:55 PM

they said nothing about possible interactions with these so called clouds, how will that program be written. it would have to be different for every single thing unless they want vines acting like rocks

8/2/2011 12:19:15 PM

Why would interactions be a problem? I know nothing about this, but if the cloud posed a problem why couldn't you model collisions with a low polygon model?

8/2/2011 12:32:35 PM

Because the CPU would need to identify exactly which if the trillions of particles is being reacted with. With polygonal, you just have to intersect a ray with a plane.

And rigging. Jesus Christ, rigging. I'd imagine you'll just end up grouping clouds of particles in a collection similar to a polygon for rigging anyway. No way you're going to do procedural animations of clouds (of significant particle depth) in real time with dynamic interactivity.

[Edited on August 2, 2011 at 12:59 PM. Reason : .]

8/2/2011 12:57:44 PM

So what exactly have they figured out to put them so far ahead of everyone else?

Don't know shit about this, but find it hard to believe what appears to be a relatively small company has gotten so much further ahead then anyone else in the industry who is pumping money into this. Seems interesting and obviously cool if it works out.

8/2/2011 1:12:37 PM

^^ there are known heuristics to solve that problem...

Quote :

" Also, if I have to hear how they're not artists and not a game company one more time....would it have killed them to hire an artist? If their engine is as the shit as they say it is surely they could find some investors. "

Because there engine can't do shadows yet, you can only polish a turd so much.

8/2/2011 1:21:35 PM

The comparison between crysis and their demo is not really fair. Crysis is a fully-developed game and not just created to show a palm tree. As pointed out above, the interaction with these 'atoms' is what makes this seem impossible. It's one thing to display a grain of dirt. It's another to display how all these grains interact together with a foot kicking them up.

8/2/2011 1:21:45 PM

I still don't get it, guys. Take their scanned rock with supposedly a shitton of atoms. Why the hell do you have to base your collision/interaction off of it's cloud?

Why not create a crude polygonal model with a few dozen polygons to simulate physical interaction and call it a day? The cloud only seems relevant for aesthetics. What am I missing?

8/2/2011 1:26:42 PM

Oh man I can't want for multiplayer online porn sim games.

8/2/2011 2:59:21 PM

http://notch.tumblr.com/post/8386977075/its-a-scam

I respect Notch's opinion, but I guess the rest of you are just going to be like SHITTY JAVA CODER RAWR RAWR.

He mentions two big weaknesses:
-Little but a handful of structures repeated (which is obvious from the video)
-Poor for animation.

I don't see anything about interaction.

8/2/2011 3:01:13 PM

I haven't heard about Voxels since Command and Conquer: Tiberium Sun was coming out back in 99

8/2/2011 4:07:48 PM

i love their amps

8/2/2011 4:22:09 PM

jbtilley, I heard that Diablo III is just like that

8/3/2011 12:56:05 AM

Quote :

"He mentions two big weaknesses: -Little but a handful of structures repeated (which is obvious from the video) -Poor for animation. I don't see anything about interaction."

To be fair, notch was a little overcritical with his post.
Even in modern games (like Crysis) there is MASSIVE model reuse. So it's a bit hypocritical to claim that as a "scam". And there's tradeoffs in storage for both approaches.

Yes point clouds take more space than poly models. But they don't need environment or bump maps, and they don't need the usual 2-3 lower poly distance models.

The animation one is a very fair point, but there's no real reason why you couldn't use a poly rigging system with point clouds. It's a technical challenge but not impossible.

Lighting is going to be a bitch, but it looks like they've got something working there, and honestly even modern games achieve lighting through serious "hack" methods. If you aren't raytracing, you aren't doing real lightings.

8/3/2011 1:19:54 AM

It may not be a scam or even vaporware, but it seems completely without application, and I doubt we'll be seeing much else from this company. I'm not exactly a computer expert, but I know enough to be skeptical.

I guess we'll see if their second demo ever shows up, and if they actually talk more than hot air in it.

8/3/2011 10:09:24 AM

Quote :

"I guess we'll see if their second demo ever shows up, and if they actually talk more than hot air in it."

Well, supposedly this is their second demo. They've been working on it for a year. If it's vaporware, that sure seems like a lot of wasted time.

8/3/2011 10:12:53 AM

All they want is to be hired on somewhere else for bank.

8/3/2011 12:29:22 PM

Quote :

"The animation one is a very fair point, but there's no real reason why you couldn't use a poly rigging system with point clouds. It's a technical challenge but not impossible. "

If lighting, animation, and collision detection are issues it seems like one way they could implement this is to only use it to increase the draw distance. Just have everything in the immediate area "normal" polygons out the wazoo, then make everything at a reasonable distance atoms or whatever.

If it really is so much less resource intensive maybe they could limit its scope to make infinite draw distance - no more fog or pop-ins.

[Edited on August 3, 2011 at 12:55 PM. Reason : -]

8/3/2011 12:53:51 PM

Quote :

"Well, supposedly this is their second demo. They've been working on it for a year. If it's vaporware, that sure seems like a lot of wasted time."

Vaporware inherently necessitates that illusion of wasted time. Seen projects just as extensive (though not as potentially ground-breaking) go on for over five years and then just fall off the face of the earth.

8/3/2011 9:08:56 PM