Can Games Look Real?

[Aidan]

http://www.gameinformer.com/b/news/archive/2011/08/02/can-games-look-real-today.aspx

[SkyLine]

1. I heard that was fake (no idea if thats true or not.)
2. That doesn't even look good. It looks worse than Unreal 3.
3. Meh.

[=CJW=YalK]

Cool

I read that it was real, just not quite up too what they claim...

Still, if it's real....bravo and yay.....if fake, then oh well...I'm not hooked on graphics anyway

[=KoS= Saber15]

All the mindless Notch worshipers spamming the company's video are extremely annoying. http://www.facepunch.com/threads/1113242-Notch-vs-Unlimited-Detail

I don't think they'll replace polygons for animated objects, but I'd love to have this for static world objects.

[Eviticus]

2. That doesn't even look good. It looks worse than Unreal 3.

To quote the video, "We're far from the world's best artists." The video is meant to showcase hardware potential, not how pretty they can make something. This video is meant to impress gamers, true, but I think they went about it the wrong way. Too many are only looking and not listening. I'm no modeler, but from what I've seen one of the issues with newly made mechs(out of the several problems that face the staff) is polygon count. Got to have enough that it looks like a mech and not a giant cardboard box with a pretty mesh, but gotta have few enough that it doesn't try to kill the game running such a complex device.

In short, if their breakthrough is legit, don't look to them for graphical wonders. That's the artist's job.

Just a few seconds before the link above you can see this, a Crysis 2 tree. Do not look at it as far as how the bark looks, that is simply the image pasted on to the in game geometry that is the 'tree'. Instead look at the shape of the tree. Then compare it to the shape of the tree shown with their system, can you not see the difference in the shape? That is the point here. Not how pretty it looks but how real the geometry looks.

[=KoS= Tripod]

wat e-man said

The pebbles, man, Look at teh pebbles! I always wanted those

[SkyLine]

2. That doesn't even look good. It looks worse than Unreal 3.

To quote the video, "We're far from the world's best artists." The video is meant to showcase hardware potential, not how pretty they can make something. This video is meant to impress gamers, true, but I think they went about it the wrong way. Too many are only looking and not listening. I'm no modeler, but from what I've seen one of the issues with newly made mechs(out of the several problems that face the staff) is polygon count. Got to have enough that it looks like a mech and not a giant cardboard box with a pretty mesh, but gotta have few enough that it doesn't try to kill the game running such a complex device.

In short, if their breakthrough is legit, don't look to them for graphical wonders. That's the artist's job.

Just a few seconds before the link above you can see this, a Crysis 2 tree. Do not look at it as far as how the bark looks, that is simply the image pasted on to the in game geometry that is the 'tree'. Instead look at the shape of the tree. Then compare it to the shape of the tree shown with their system, can you not see the difference in the shape? That is the point here. Not how pretty it looks but how real the geometry looks.

Interesting.

[thEClaw]

Is that like Atomontage? I hope they don't compete with each other, i would like to see a voxel-based game. And with all these companies around that probably will stick to what we know and have now, a competition might destroy all hopes for this kind of thing.

[Spooky]

It's not a Voxel based engine, it's a Point based engine ("Surfel"). And I think Notch's assessment, that it's based on sparse voxel trees is incorrect also.

[Ghiest]

But his analysis of the data needing to be calculated is correct, unless the do some funky things with occluded points it's just too much data to process.

[[IPA] Thalamus]

Impressive proof of concept, meh tech demo. And the speaker reminded me of the radio shows from Grand Theft Auto.

[Spooky]

But his analysis of the data needing to be calculated is correct, unless the do some funky things with occluded points it's just too much data to process.

Yes, you need an enormous amount of data. But the key here is, that they claim to be able to search through all the points so efficiently, that they can fetch the minimum amount of points for each pixel needed on the screen, in the correct detail level and order, at interactive frame rates.

[Ghiest]

But his analysis of the data needing to be calculated is correct, unless the do some funky things with occluded points it's just too much data to process.

Yes, you need an enormous amount of data. But the key here is, that they claim to be able to search through all the points so efficiently, that they can fetch the minimum amount of points for each pixel needed on the screen, in the correct detail level and order, at interactive frame rates.

Pie in the sky, it's a perpetual motion machine ... ill eat my words when and IF it ever sees the light of day.   Also it looks crap not because they are shit artists (I mean who wouldn't hire a freelance for a few days to produce some decent models at a few hundred dollars to showcase an engine?) but there is total lack of proper shadowing, specular, sss and any other sort of normal lighting set-up there and that shit is basic for engines.  Btw they were showing almost the same engine around 5 years ago.

[Spooky]

Btw they were showing almost the same engine around 5 years ago.

Their first video presentation was in February 2010.

[Bill]

I don't know spooky, I seem to remember seeing something somewhat similar around the same sort of time frame that Ghiest's talking about.

It wasn't a presentation per se, but it was featured on a UK based pc gaming/tech show on satelite tv and the key principles seem to be the same.