Roll on unlimited detail gaming!

[fatality_uk]

http://www.reghardware.com/2011/08/03/game_graphics_could_be_10000_times_better/

[forrestcupp]

That's flippin awesome. I can't wait to see it with multiple levels of shading.

[Erik1984]

Don't forget that so far it's only a claim and nothing real (software we can download and run to see this supposedly great engine at work) has been released.

[Lucradia]

Don't forget that so far it's only a claim and nothing real (software we can download and run to see this supposedly great engine at work) has been released.

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

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

[Copper Bezel]

It could still be used, theoretically, to build backgrounds in video games, so long as you don't need them to be interactive and can fake shadows. Or, alternately, to make explorable static models at high detail for educational purposes, like, say, of sculptures or something.

[NovaAesa]

Oh noes... it's this "unlimited detail" thing coming up again... *serious face*

[fatality_uk]

Oh noes... it's this "unlimited detail" thing coming up again... *serious face*

Oh noes... it's this how dare someone start a thread I clearly discredited on my blog somewhere in cyberspace *serious face*





;)

[forrestcupp]

I'm not going to rule it out yet. They plainly said in their video that their method isn't original, and it has been used in medicine, etc. Their claim is that they have figured out a way to make it fast so that it can be used in games. The guy who posted the blog against it was wrong about the ability to animate using voxels.

Since it's not causing me any harm, I'm willing to wait and see what further offering they bring out.

[Copper Bezel]

It's true that you can animate with voxels, sure. That's, generally speaking, what they're for. But it's a lot more calculation than bending a polygon or spline mesh, and it's hard to imagine how you could animate using this search method of rendering - or, rather, you'd start to lose the advantages once you're having to recalculate the matrix of points out there every frame because something. Add those things together, and all that instantaneousness looks a little weak.

More importantly, nothing is animated in the demo videos. = P

[vehemoth]

You can only wait and see whether it's fake or not. Has anybody got any guesses as to how this works because what I know about voxels I don't see how they could of done it.
How will this affect collision detection etc. I think in all the videos it just shows rendering and no game like or physics simulation.

[Lucradia]

I'm not going to rule it out yet. They plainly said in their video that their method isn't original, and it has been used in medicine, etc. Their claim is that they have figured out a way to make it fast so that it can be used in games. The guy who posted the blog against it was wrong about the ability to animate using voxels.

Since it's not causing me any harm, I'm willing to wait and see what further offering they bring out.

I'm guessing they just recorded demos from the GTX 600 series, which uses dynamic tessellation, which allows for "almost" infinite and different configurations of landscapes in all directions (which would be good to latch onto for Minecraft.)

[PhillyPhil]

There are no completely intractable problems with this. Animation is possible, rotation is possible, space considerations can be overcome with procedural generation and clever repetition.

It seems to me that this is a fairly obvious and logical progression in 3d graphics, and that it's only a matter of time (and processing power) before something similar to this is ordinary, taken-for-granted technology that we all use on our personal computers (whether this particular engine from this group goes anywhere is a separate issue).

[del_diablo]

1. Its a scam
2. They publish a proper tech demo, as in a executable and 30-40mb of gfx data, and show us its not a scam

[PhillyPhil]

I think people are (mistakenly) condemning the technique along with the implementation.

As a proof of concept, this example is sound. There is no reason to think that we won't see graphics done this way in our games in the not too distant future.

Is this particular engine, from this particular entity any good? I have no idea, but let's not throw the baby out with the bathwater.

[Lucradia]

I think people are (mistakenly) condemning the technique along with the implementation.

As a proof of concept, this example is sound. There is no reason to think that we won't see graphics done this way in our games in the distant future.

Is this particular engine, from this particular entity any good? I have no idea, but let's not throw the baby out with the bathwater.

Fixed; because we need larger harddrives, like, think over more than 50 TB to handle all this. Also more RAM. Motherboards aren't advancing fast enough for this to happen anytime soon.

[PhillyPhil]

Fixed; because we need larger harddrives, like, think over more than 50 TB to handle all this. Also more RAM. Motherboards aren't advancing fast enough for this to happen anytime soon.

50TB? Not so. Procedural generation and repetition will be how the data size problem is solved.
Motherboards aren't being 'held back' for any reason other than a lack of reason to advance faster.

[mips]

silicon snake oil

[forrestcupp]

It's true that you can animate with voxels, sure. That's, generally speaking, what they're for. But it's a lot more calculation than bending a polygon or spline mesh, and it's hard to imagine how you could animate using this search method of rendering - or, rather, you'd start to lose the advantages once you're having to recalculate the matrix of points out there every frame because something. Add those things together, and all that instantaneousness looks a little weak.Fortunately, in this case, the responsibility is on Euclideon to figure that out, and not me. ;)


Fixed; because we need larger harddrives, like, think over more than 50 TB to handle all this. Also more RAM. Motherboards aren't advancing fast enough for this to happen anytime soon.There used to be a game on the Commodore 64 called Outrun, which was a port of the arcade racing game. The C64 only had 64KB of RAM, yet for its time, Outrun was a surprisingly beautiful game that appeared to be 3D. They used tricks to get that effect, such as reusing objects and scenery on the side of the road with the understanding that the player is driving the car so fast that it wouldn't be noticeable. They did all of that in 64K and for its time, it looked beautiful.

With our current limitations, there's no way we could have that many distinct "atoms". But I don't see why they couldn't use tricks to reuse a lot of information for things that don't matter that much, like blades of grass and other scenery. They already do that with polygons.

I'm willing to wait and see if they come out with any more evidence.

[Lucradia]

Do they have a less-than 20 MB Demo file that we can download and run?

[forrestcupp]

Do they have a less-than 20 MB Demo file that we can download and run?

Even an optimist on the matter, like me, wouldn't require them to fit all that in 20 MB.

I'm not saying it's not fake. I'm just saying I hope it's not, and since it's not harming me, I'll wait around and see what they come out with next. ;)

[JDShu]

More interestingly, they spent a year in hiding and came out with another video demo. That's just too slow regardless of whether they have something real.

[del_diablo]

Fixed; because we need larger harddrives, like, think over more than 50 TB to handle all this. Also more RAM. Motherboards aren't advancing fast enough for this to happen anytime soon.

Take a look at the latest rage demo: 1 terabyte of data, IF it was uncompressed.
This is insane and stuff. Why is it insane? Because 90% of that data is just "handdrawn stuff, saved as harddrawn stuff".
1. Remove unwrapping
2. Allow to assign a material to each face
3. Allow the material to be procudarely stored
And suddenly Rage is perhaps... 100-500mb of compressed data?
The same principal applies to any game.

Or lets take a a old Nes game, perhaps something like Metroid? If you had saved that, pixel by pixel, as uncompressed data, there would be a lot of data for its time. It is equally ridiculess as your 50TB claim. Nobody stores data uncompressed, or in stupid ways. You store data as metadata, and you compress that metadata.

[Lucradia]

Take a look at the latest rage demo: 1 terabyte of data, IF it was uncompressed.
This is insane and stuff. Why is it insane? Because 90% of that data is just "handdrawn stuff, saved as harddrawn stuff".
1. Remove unwrapping
2. Allow to assign a material to each face
3. Allow the material to be procudarely stored
And suddenly Rage is perhaps... 100-500mb of compressed data?
The same principal applies to any game.

Or lets take a a old Nes game, perhaps something like Metroid? If you had saved that, pixel by pixel, as uncompressed data, there would be a lot of data for its time. It is equally ridiculess as your 50TB claim. Nobody stores data uncompressed, or in stupid ways. You store data as metadata, and you compress that metadata.

Then make them release a 500mb demo.

[PhillyPhil]

Then make them release a 500mb demo.

Why should it be 500MB? You're just pulling figures out of the air.

The point is that even if this individual implementation is less than useful (something we don't know yet) the concepts behind it are sound.

[Lucradia]

Why should it be 500MB? You're just pulling figures out of the air.

The point is that even if this individual implementation is less than useful (something we don't know yet) the concepts behind it are sound.

We know the concepts are sound. It's like the GTX 560+ Tessellation system.

[PhillyPhil]

We know the concepts are sound. It's like the GTX 560+ Tessellation system.

I wasn't replying to ''we'', I was replying to you. And no, this is not tessellation.