Discussion in 'Article Discussion' started by CardJoe, 2 Aug 2011.
SOUNDS LEGIT. Give me a tech demo that I can run on my PC and I might believe it, they've shown nothing so far that can't be faked.
We already have a discussion going in Hardware about this, just FYI
My only problem with this video is the awful voice over.
100,000 is a pretty big numbah..
"They depend on stancing because every distinct model requires several dozen MB of storage. This is and has always been the biggest drawback to voxel rendering. You might be able to "find" the correct voxel in an blink, and you might only need find 1920x1080 voxels to make the picture, but the billions of voxels required to represent a voxel made scene decently still have to be stored somewhere.
So... they cheat. They put the same models over and over and generate some things procedurally, which is not a bad thing to do on it's own, but is not related with their tech at all and it's just bypassing the real problem with voxel rendering. Using procedural methods, you could generate the same detail and randomness with polys, and definately when applying tesselation. What you can't do that way is to create the complex, unique snd varied scenery that we are used to see in modern games.
Quite literally everyone in the gaming industry is researching/has researched voxel rendering and literally all who did have abandoned it or delayed it to 2016 ++. Unlimited Detail is just a hoax or a scam. They are doing voxel rendering and doing it pretty fast according to them... well I'm not even going to try and refute that claim, there's over 50 other voxel renderers out there capable of similar performance (though most use the GPU), already with proper real time demos and proper papers and also patents. None of which UD has shown. It doesn't matter, none of the others are suitable for real time today and neither is UD. "
Posted it in the Forum this morning xD keep up people xD.
Well maybe it’s just the child inside or me not thinking rationally but I really hope that this idea comes to fruition as it would revolutionise the gaming industry. Whatever artists really what to create in terms of their visions will no longer be hampered by the technical restrictions anymore.
That voice-over guy is awesome - I wonder what would happen if you just gave him a LemSip?! Interesting technology, although the major reservation I have is the amount of identical objects shown off in the demo. They may indeed show off hundreds of thousands of 'atoms', but this isn't really a big deal if they're all the same - you just instance loads of the same model. If they can do the same thing with lots of different-looking models, then that would really be something.
So it is Voxels then. First thing I thought when I read this. Why aren't they calling them voxels?
+1 to that!
I really hope this is legit because all the flaws the video has pointed out with poly rendering have annoyed me for a long time now. Also if decent physics are implemented into the engine that would be great.
whats that saying, the one about being too good to be true? ....
Note that everything that they've shown so far has been entirely static. Nothing moves (though they're careful to always keep the camera in constant motion to make it less obvious). Not too great for a game engine.
I've seen real-time-rendered voxel engines with on-the-fly geometry modification (e.g. Atomontage) but they're usually of a lower resolution. To have both at once, you need faster processors and MUCH MUCH more storage bandwidth.
I think they are working on the wrong thing here. Polygon detail is not the major problem facing most games. From a static scene point of view the biggest improvements seem to be with lighting these days. That's not to say a few more polygons wouldn't be good, but there are lots of ways around that - normal mapping, displacement mapping, tessellation.
However the current challenges are more to do making stuff move and interact better. e.g. destructible environments, and animation (realistic faces and body movement).
Anyone else notice there were exactly no shadows in any of the test videos?
I'd love to see how they calculate how a billion voxels would interact with a dozen moving light sources.
This is the perfect logical next step in CG, Polygons have had their days numbered for a while now. If these guys get it right, they could become mega rich
Maybe I'm in a minority, but I want my games to look like computer games, not like real life. I think the difficulty is that something that looks almost but not quite 'real' is more disturbing (and disappointing) that something that is clearly made up.
+1 he does sound like your typical voice over for parody videos... in my opinion.
What are your credentials? I mean you say it doesn't work and we should believe you why? I always question everything. I will believe UD when I see it in a modern game but I don't automatically dismiss it because you come along and say it doesn't work.
I agree, to me the majority of games I play are for me to "escape" reality and indulge in something different. Sure a realistic game could be graphically enjoyable but I won't get the same escapism factor.
So I'm interested, and of course we aren't going to see anything out of it for... well 2 years optimistically. I'm interested in how they do the textures, are they simply coloring each 'atom'/voxel object to match where a texture would normally be plastered over? I'm also interested in 'procedural destruction', I guess it would be called, where these hollow voxel object can be cut or broken and fill in the pieces, that would be a major improvement over the "Cover in smoke/replace with broken model" approach to destruction.
According to the video description they have animation in the works, so if they show that and their improved shader off in the next month, I'd be very impressed. Drop a Physics engine on that and they probably will get quite a few sales.
Separate names with a comma.