News Euclideon promises unlimited graphical detail

It'd be interesting to see if this technology could perhaps be intregrated with polygons. Things which look particularly bad and are easily repeatable like round trash cans in a city street could use voxels to smooth them out. No need to make them move, simplistic lighting, perfectly acceptable to re-use the same model over and over.

 

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

Notch has spoken

 

Just what I was thinking! I'm still sceptical, but if true, wow...

Concerning the (complementary) side issue that's popped up about the level of realism in game graphics, I think I'm in the middle of the argument: Some games I'm happy to become as realistic over time as they can make them - and if they can eventually make them indistinguishable from reality then brilliant. Other games though, while I'm happy for them to improve on where we are now, I think there are limits as to what I would like to see. For example:

If it's a driving game, flying sim, puzzle-type game or maybe something set in space - make that as real as you can make it, I'd love that. However, if I'm going around stabbing people with broadswords, smashing their skulls in with battle-hammers, shooting them in the face with rifles etc, etc - there are limits on how real I want that to look. I'd like it to look good, but I want it to remain instantly and obviously recognisable as artificial.

As CPU's, GPU's, APU's & the software that power them get ever more powerful, I think this is going to become a really, really big issue...

 

Very much agreed .

 

I played Just Cause 2 for over 17hrs at 6000x1080p and never noticed those crappy sprite leaves in the trees. At least I don't remember noticing them, maybe I did, rolled my eyes then went back to enjoying the hell out of the game.

I'm all for huge leaps in tech but if the realism part doesn't start catching up things get weird. I think resident evil 5 is a perfect example. The game looks freaking awesome but when you can't just duck under a laser it kinda ruins things. Especially whenever you are a freaking super hero during cutscenes!

Glad too hear notch's take on it as well.

 

So much this! Sometimes things don't need to realistic, just well presented. I mean look at Morrowind, it's as realistic as...wait it's not even realisitic. But it's still incredibly immersive.

 

Played Killing Floor a couple nights ago, didn't notice the Unreal Engine 2 graphics at all. Got so wrapped up in the game that it just didn't matter.

The graphics are just the icing on the cake, you've got to have some icing to make it look finished and presentable, but no one's going to want to eat it if you spend all your time decorating and no time adding flavor.

 

I've seen this before. I don't remember when, but I swear I've seen this concept before. About 5 years ago? I'm sure some of you with a clearer memory than mine has seen this as well.

 

*Search Internet for "Sparse Voxel Octrees"*
*Learn about "Sparse Voxel Octrees"*

Because the sooner you do this, the sooner you realise this "techdemo" is nothing special. Graphics of that type have been around for years, so have voxels. The test is applying them in a meaningful manner.

Voxel's PRIMARY FLAW: They are extremely difficult to animate smoothly
The model is designed in "static" form and as soon as you try to animate this model, its static form is no longer what it was. Arms get longer, skin stretches. With polygons this is not an issue, stretch them halfway across a map if you like, not a problem. How ever, since voxels are tiny little 3d pixels, as soon as you start animating a voxel model holes and tearing start to form where the voxels no longer fit flush.

Carmack talked for a while in his 2010 (i think) keynote speech @ Quakecon. He discussed the use of a sparse voxel octree to render static geometery. Terrain, buildings, etc, etc. It would allow "infinite" detail in the same way Megatexture allows infinite detail. It will stream the detail as needed from the harddisk.

During this same talk he also discussed the future of raytracing vs rasterization. His opinion is that one is not "better" than the other and for the best payoff between looks and performance hybrid systems should be considered.

 

Let's put it this way, if you are simulating something then you want the graphics to be as realistic as possible along with the physics. I don't think advancing graphical realism takes anything away from games. I'd love to see a horror game with hollywood level special effects just like I want to see photorealism in something like Battlefield. There is no line that needs to be drawn since as hardware advances, so we have the potential to create more cinematic game experiences.

 

I wasn't putting the advancement of graphical potential down at all, I game more than regularly and have put 450+ hours into Bad Company 2 on the PC. I have been playing Battlefield games on the PC since 1942 and have loved seeing the games develop over time, I was recently on the Battlefield 3 Alpha for the full duration of it too (not the 1 weekend bunch) and seeing the Frostbite Engine in its next stage was pretty impressive (even though it was not at release quality).

I don't want to for one second to stop the progression of tech, especially gaming.

 

I have a lot of catching up to do then on BC2 because I haven't played it much this year. I completed the single player then online here and there. I do like it a lot when I play it though. I don't have enough hours on BC2 to qualify for the BF3 alpha.

 

interesting

 

There will be an open beta in September for anyone and everyone to get their nose into the game, I look forward to seeing how it has progressed in the coming few weeks too

 

Wonder how much theyre looking to sell this new wonder technology for. Sound like a bunch of get rich quick cowboys to be honest.

 

Why don't they throw some realistic water effects and advanced lighting, and maybe wind. Tell you what, throw in some explosions and moving characters. Lets see how well they can pan and zoom around the map then.

Don't get me wrong, I'll welcome it if if can be incorporated into a game and still run on a medium/highend system. It just wasn't very convincing for me.

 

the atomage demo uses polygons for the vehicle and voxel's for the terrain. There's no need for an engine to be all one or the other and I think the strengths and weaknesses would play right into the hands of each other. Most terrain and assets are static or non-deformable. They would probably benefit from voxel technology. It seems there might be programming gaps before it could be used for fluid character animation. The perfect example is those trees or other assets that could be instanced many times and are static. Same for most terrain or buildings. If voxel technology is so great than you could probably increase the level of detail a few times where you could use the voxels and increase your polygon budget on animated models.

I don't think the technology is a scam I think the strengths and weaknesses need to be understood. That being said the atomage techdemo had a very interesting artistic style and would be awesome using the larger voxels as is.

 

Imagine the damage models with this tech, throw grenade, bits of the ground go away...

 

Also I think notch is off base, he makes the point that it wouldn't be possible to do assets, assuming no compression, no instancing and that full resolution was maintained in the entire volume when in reality it's all about the spare voxel techniques. I think it would be quite possible for some procedural techniques to maintain high resolution with low total voxel count and end up at very reasonable file sizes