New Graphics Technology?

[seaeye]

Came accross this while clicking through a Minecraft forum..

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

Is this genuine?

[CharveL]

Seems plausible to me. There might be some difficulties in fast ram storage and access for the point cloud to reside, or applying shader type effects I could see. Maybe just small hurdles really but I don't see why we shouldn't be optimistic with this approach. It probably depends on their method or hardware requirements for fetching the proper pixels to display but even their realtime demo is very impressive at this stage.

Look for the interview they did where they let the interviewer zoom around in the island. Right down to individual pebbles on the paths and fully 3D leaves, so it seems to work.

[Buchon]

I wonder how they will going to calculate collisions or animations with that amount of points.

[ATAG_Doc]

quantum leaps examples
http://sanfrancisco.cbslocal.com/2011/08/26/google-brings-ultra-fast-internet-to-homes-near-stanford/

and

http://www.msnbc.msn.com/id/3077363/ns/technology_and_science-science/t/quantum-leap-computing/

I think we might all be surprised to learn that computing is on the verge of a major leap once again. But the truth is the shares of PC's as in terms of people owning them and total sales as a percentage of all computing devices has really shrunk and have been on a decline for a while. Which is sad.

Looking at that article from 2000 makes someone like me who is old enough to remember 300 baud modems and bulletin boards be in awe at where we are today.

[DayGlow]

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

Interesting commentary on the engine. I'll believe it when I see a decent working demo with transparency and animations.

[Madfish]

Definately not a scam. The engine just has limitations. Those can be overcome however.
For example a mix of both worlds. It was done already by many games like Outcast or the Delta Force game series.

If applied correctly this technology could seriously enhance games. Even decade old games like Outcast are still visually impressive and were long unrivaled.

[albx]

The video is a pre-rendered scene

[CharveL]

The video is a pre-rendered scene

Like I said, youtube up the realtime video.

[CharveL]

I wonder how they will going to calculate collisions or animations with that amount of points.

Well even CloD needs separate damage maps but I'm thinking that simplified underlying polygon structures will do for collisions considering all the extra GPU headroom.

Without ray-tracing I can see difficulties with proper shadows and other visual shader effects but there is currently so much waste with having to calculate hidden polygons and overdraw despite some things they are doing with occlusion and tessellation.

[CharveL]

http://www.youtube.com/watch?v=JVB1ayT6Fdc

[Jatta Raso]

http://www.youtube.com/watch?v=JVB1ayT6Fdc

i've hold my judgement until he starts introducing his team. after that, well, what can i say except WHAT A HOAX!!

seriously, weighting all in all, if these ppl are into building revolutionary graphics engine then i am Donald duck; you should definitely recognize a scam like this when you see it, this one it could be a compendium of scam tricks

[Warhound]

http://notch.tumblr.com/post/8423008802/but-notch-its-not-a-scam

Followup post by Notch elaborating exactly why it's pure marketing talk by guys looking for nice fat cheques. Certainly nothing you'll see in games anytime soon.

Here's an example of something voxel based that actually does exite me. :)
http://www.youtube.com/watch?v=_CCZIBDt1uM

[Madfish]

http://notch.tumblr.com/post/8423008802/but-notch-its-not-a-scam

Followup post by Notch elaborating exactly why he believes it's pure marketing talk by guys looking for nice fat cheques. Certainly nothing you'll see in games anytime soon.

Fixed for ya. :P

Like I said there are tons of games out there that use this or similar technology already.

* He was wrong about point clouds.

* He's riding the voxel train although this technology is freaking old and, like I pointed out, produced fabulous games already.
http://en.wikipedia.org/wiki/Outcast_(video_game)#Graphics_engine and http://en.wikipedia.org/wiki/Delta_Force_2#Game_engine for example

* He assumes that the new engine would do EVERYTHING in point clouds, which is rubbish. It's like saying current gen engines do everything with polygons, which they don't.

* He talks about how they do false advertising - which they don't. Of course atom based rendering is kind of a new technology, or at least one that was revived against all odds, and of course many are researching it - it's still not what we currently have and thus it's new.

* He also says he's wrong in his own words: * They seem to be doing some very impressive voxel rendering stuff, which could absolutely be used to make very interesting games, but it’s not as great as they claim it is.
Claims aside, and don't even get me started about his game, minecraft, which has one of the ugliest graphics engines PC gaming has seen in decades, they WILL have to prove to the industry that their engine IS capable. They're not a game studio like his - they're just a tech company. I doubt they'll EVER develop a game.

* Furthermore he goes on about hardware - well, yes, hardware. Everyone knows that computer games are being developed with the FUTURE in mind. So if these guys come up with a reasonable product within the next 3 years then it might take 8 years before we see first applications! That is a long long time and I would say we could definately see raytracing, point clouds and reborn voxel technology by then.


Just think about how the new Tech5 engine from ID revolutionizes the market and that just by "megatexturing (http://en.wikipedia.org/wiki/Megatexture)" which was also thought to be impossible back in the days.

So yeah, stop being so pessimistic. We won't see it within a couple of years, that's for certain but I don't think it's a scam. And if it's funding fraud you're so worried about, talk to your government - they're experts on it on a MUCH larger scale. ;)

[CharveL]

Why pick sides? It's an interesting concept that's bound to have some value and more than likely they'll get bought up by a bigger tech company.

The big guys know how to evaluate the worth of potentially useful IP so who knows.

[Madfish]

Not picking sides. The only side I pick is the one saying that this is not magic or a scam.
There are a number of engines in development that use new approaches to the old voxel technology or are going for atomic / point clouds. I couldn't care less which technology sees the light of the day first but it isn't a scam and most certainly not fake or a lie.

Since it's worded weird again: I'm not defending THIS company. In fact they way the present it is surprisingly arrogant and suspicious.
Couple things I find really questionable:
*The guy doesn't know what LOD is / means (not level of distance but level of detail)
*He messes up where tesselation was mentioned or rather what engines / games have it implemented. Even if he's doing a different tech he should know about such basic things that every gamer kid could tell you about.
*He messes up often when it comes to numbers - they're rarely precise - uncommon for an engineer.
*Weird presentation of the "team" at the end, even containing advertisement for a flower shop.
and other stuff. But I still hope that real investors would test it thouroughly anyways. If I were a big games company I'd go there and have them show me before putting millions on the table.
So there's suspicion and hope. ;)


Of course we won't see it tomorrow - not even with a billion dollar investment. Both hard and software need to grow a bit more. But saying "notch" said is just weird. Especially because "notch" has a few mistakes in his posts and also because "notch" needs to clean his own backyard first. Minecraft is surely a game in a questionable state but that's not this threads topic. :p


One thing is certain though. Newer games require such massive amounts of content that the classical polygon approach is simply impossible to maintain. For example modern shooters with destructible houses etc. - you simply can't afford those hordes of people crafting polygon models for all different kinds of destruction etc. anymore. Things need to go from "handmade" towards a more automatic approach through defined materials that respond naturally.


If we're lucky the next IL-2 in 10 years uses such technology. An engine that'd let us do ermegency landings in the sand of african deserts. We could see the wheels digging into the sand then being ripped off along with the gear, being left behind on the ground while the wings slice into it and cause a rain of dust, leaving some of it settling down on the bent fuselage. The canope opens with a cranky sound and wind blows sand onto the instrument panel... ;)

I'd say let's hope they get things going quickly :cool:

[CharveL]

Funny but I don't really see the "arrogance" with this guy that others do. He's clearly not a marketing trained guy so he doesn't come across very well but that doesn't have any bearing on the technology he's pimping whether good or not so good. I have to admit I was saying to myself, "wtf??" when he used the LOD acronym wrong.

But short of some elaborate conspiracy hoax where the interviewer was in on it and they played some well-timed, pre-rendered movie while pretending to move it around with a controller, I think it's pretty clear that the engine works. Who knows if there's some major brick wall limitation lurking behind the scenes that makes it completely useless for games.

To me, it seems like there is a huge database storing each cloud point atom's position and attributes. The engine has to:

a) determine by means of controller movement and position of camera on the map, which scene will be displayed next.

b) Then go fetch say 1280x720 pixels every 1/30th (or whatever speed it can) of a second.

c) display it on the screen.

By not having to determine and process visible triangles in a scene, many of which will not be seen, it should be more efficient in theory. This is a vast oversimplification because I can see where it would get difficult determining, say, what one screen pixel should look like when, at a distance, there are 600 pixels in that space. Or how do you go about deforming and applying physics to the world beyond the few canned animations in the video (grass waving). Wouldn't it work at least as well as voxels? Unless the additional detail/fidelity brings physics calculations to the computer's knees having to process 64 pixels per millimeter.

But still, it seems that this is achievable, especially once they start tapping the GPU. At the very least I can see Microsoft scooping them up to use in their Photosynth technology before it starts getting used in games.

Apparently they aren't even asking for anymore money right now as they have enough to get to the next stage, so just guessing that it's a scam as a reactionary statement to set oneself up to look smart if it turns out that way, why not just take a look at what's there with interest and hold off on sending them a cheque they don't even want?

Funny, I've been following graphics since the original Doom/Duke Nukem days with interest and must have missed this Notch guy. Although I'm sure he's very talented and his game a load of fun, I can't really recall him being any expert in all things graphics. Carmack on the other hand should know his stuff but his time is long passed as far as cutting edge relevance afaik.

Meanwhile I'm going to stay interested and healthily skeptical, looking forward to what they manage to come up with the next time they come out of hibernation.

[CharveL]

http://notch.tumblr.com/post/8423008802/but-notch-its-not-a-scam

Followup post by Notch elaborating exactly why it's pure marketing talk by guys looking for nice fat cheques. Certainly nothing you'll see in games anytime soon.

Here's an example of something voxel based that actually does exite me. :)
http://www.youtube.com/watch?v=_CCZIBDt1uM

That would excite me too if it could be done on something larger than a postage stamp. ;) Still, even at an early stage there is a whole new field of gameplay possibilities using voxels or this unlimited detail type engines that don't require huge maps in the first few years until hardware opens it up more.

The money saved from studios tweaking polygon counts and such can be redirected to making storylines less hokey and vapid like they are in 98% of releases now.

[Madfish]

He's arrogant because he's playing down the arguments others have which are reasonable in some cases. This engine has definite limits and he isn't mentioning them at all - no, he's also lieing about them. Knowingly or unknowingly.

The most obvious argument however is that he says "infinite". Infinity is impossible, not even the universe itself is infinite. That's pure marketing talk and it IS either stupid or arrogant.


Notch isn't a known guy at all. He's just very well known with the kids since he is the guy who made Minecraft. An indie sandbox game that had a huge success in terms of sale and media attention. Like a virtual lego.
Ironically it has a lot of fundamental physics and logical issues too though. That's why I said he needs to clean his own backyard first.
Also the game was a crazy seller and then development stopped. It's literally much worse than Cliffs of Dover if you take a look at how much money was made.
Further, although the game is still alpha or beta (yes, it was sold as an alpha) they are working on a follow up game and the development for the original title is basically dead.

So overall I don't particulary like Notch. In many ways. But like there is truth to both sides there are also lies or false statements.



You said you'd like an engine like that but not as small as a postage stamp. Well, I imagine the first real uses we'll see will not simulate everything in perfect detail. Rather the technology will be use partially - for example for the areas you can use and also only partly with these.


But then again, the thought of an airplane crash landing in sand is pure awesomeness. ;-P

[CharveL]

No doubt! :cool:

I just think people get so hung up on the nuance of words and perception of attitude. The guy says infinite assuming everyone knows that nothing is technically infinite, and is just trying to make the point that there isn't any theoretical limitation to the size of the point cloud they sample from.

The guy is clearly pretty nerdy and not the best with words or presentation so I give the benefit of the doubt solely on the fact that he's not some slick marketing type that says all the right things in the right way. He just sounds absolutely convinced of his own tech and where it's going (whether it actually does or not).

[raaaid]

machines cant reach infinity, maybe spiritual things can

well unless a machine is spiritual

whats this?

X

a road in perspective that goes to infinite and beyond :)

[M1sF1rE]

The rendering is not the trick here people. All rendering works the same. You turn a pixel a certain color and render the scene.

Not having to recalculate the scene "on the fly" would save a huge amount of gpu/cpu power.

Maybe every atom has an address in 3d space and the algorithm simply has to calculate which ones are visible at your current location.

How you determine where all those points are in 3d space (which you would need to know in order to determine the next frame) would be the magic part.

It seems to me the algorithm and the data compaction would be the break thru's with this engine.

Assuming it's not a hoax of course.