Voxel-based Infestation

zex

Since there are some limitations to polygon-based dynamic infestation, which are presumably delaying it's introduction into the game, I'm going to suggest a different kind of system based on voxels. There has been a ton of progress recently in the field of sparse voxel octree rendering allowing essentially infinite fractal detail in games. It is most commonly used for far-distance terrain rendering in games, but I think this technique would be ideal for Infestation in NS2.



<center><object width="450" height="356"><param name="movie" value="http://www.youtube.com/v/QsUm42DJQZI"></param><embed src="http://www.youtube.com/v/QsUm42DJQZI" type="application/x-shockwave-flash" width="450" height="356"></embed></object></center>

<center><object width="450" height="356"><param name="movie" value="http://www.youtube.com/v/1sfWYUgxGBE"></param><embed src="http://www.youtube.com/v/1sfWYUgxGBE" type="application/x-shockwave-flash" width="450" height="356"></embed></object></center>

Zuriki

I'm in, but I'm out. Voxels would be great here, but I feel that UWE probably has more experience dealing with polygons than voxels - I'm not saying they couldn't do it, just it would probably mean they have a lot more to learn before they can implement it successfully.

HadesDK

Looks great

Chris0132

Er I think you would sort of have to entirely rewrite the engine to render voxels. Rendering polygons and rendering voxels are like two entirely different approaches to rendering.

Hanneslol

I don't think it would require much actual rework of the engine, but merely addition to it. Also I suppose it would require more performance to do Marching Cubes or something similar every time it needs to update.

Vladimir Van Vodka

Sorry, but voxels are against my religion, I can't support this.

endlesschange

Looks like a tree.

Jason Whore

vawksels !
hm. isnt that kinda... ah, well forget it.
yup, sounds good.

i'd like a uwe programmers opinion on this

Rokiyo

The problem I have with the idea of growing a polygon mesh over the top of the underlying map is the same one that I have with the current decal-based approach: Every additional polygon (or decal) adds extra computational strain that wasn't there at the start of the round. You end up with a system where the more infestation a player grows, the more your performance takes a hit. You end up in territory where players start chastising each other for pursuing otherwise valid tactics on the basis of performance.

I rather imagine Max is shying away for similar reasons. You wouldn't expect the man capable of implementing a lighting system that handles 30 dynamic lights and zero dynamic lights equally well to turn around and implement infestation in a way that degrades over time.

Stuff like voxels and tessellation seem great in this regard because it looks like rendering a flat surface will be equally difficult to rendering a highly deformed one. Takes a bigger performance hit up front, but ensures the framerate remains steady throughout the game.

Personally I'm wondering about the merits of incorporating DirectX texture blending, and having infestation abilities modify the alpha maps of the textures on the level geometry itself.

Josh86

Make as much as you can in voxels. I loved Delta Force.

I don't know much at all about programming games or graphics/rendering. From what little demos and writing I've seen on voxel rendering I can't say why it's not more popular. You bring up a good point with the rendering of distant landscapes/terrain, though -- seems perfect for that.

I should read more about it. Voxels seem cool enough.

Max

<!--quoteo(post=1856259:date=Jun 25 2011, 02:30 AM:name=Hanneslol)--><div class='quotetop'>QUOTE (Hanneslol @ Jun 25 2011, 02:30 AM) <a href="index.php?act=findpost&pid=1856259"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I don't think it would require much actual rework of the engine, but merely addition to it. Also I suppose it would require more performance to do Marching Cubes or something similar every time it needs to update.<!--QuoteEnd--></div><!--QuoteEEnd-->
The original infestation demo uses marching cubes.

Josh86

I don't remember things very well, but wasn't the idea of the aliens using tunnel travel in combination with the infestation brought up before? I could see this being a really spectacular way to implement that. "Carving" tunnels through geometry and all.

I played Red Faction on the PS2 and remember it having an impressive capability for geometry destrution. Not saying NS2 needs to have destructible environments, but the idea of having map deformation sounds awesome. I admit the dynamic lighting and infestation changes rooms on their own, though.

All cool stuff. I'll be happy with UWE getting the dynamic power/lighting and infestation stuff perfected. A lot of work ahead!

zex

Yes, adding voxel growth and rendering would involve adding new systems to the game engine. But clearly adding a polygon-based marching cube implementation is also going to need further development judging from the perpetual delay. So it's just a matter of direction. The original tech demo was impressive for 2006 or whatever, but game technology has progressed to the point of fully dynamic destructable level geometry, so the "wow" factor of a marching cube mesh might not have as much impact in 2011/2012. Whereas fractal growth using voxels has never been seen before AFAIK and would blow minds for sure.

Chris0132

Because nobody has found a way to implement voxel rendering in a game engine to the point where it runs smoothly.

I know john carmack said in an interview that he'd like it for idtech six or something, but everything currently uses polygons, you'd need an entirely new set of tools to handle texturing and modelling with voxels, and entirely new rendering approaches to support it on a large scale.

I mean it would be lovely if we had it, it'd be wonderful to be able to unify textures and geometry into one system, and to not have to worry about things like LOD meshes and polycounts, and as mentioned, more detail meaning more cost, but I really don't think we're there yet. I expect that someday voxels will be the common approach to 3d environments, but generally you don't get change until everyone in the world can handle it. I mean we've had 64 bit processors for years but almost everything still uses 32 bit as standard.

Give it 10 years, maybe 20.

zex

<!--quoteo(post=1857116:date=Jun 29 2011, 12:13 PM:name=Chris0132)--><div class='quotetop'>QUOTE (Chris0132 @ Jun 29 2011, 12:13 PM) <a href="index.php?act=findpost&pid=1857116"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Because nobody has found a way to implement voxel rendering in a game engine to the point where it runs smoothly.<!--QuoteEnd--></div><!--QuoteEEnd-->

There are videos all over youtube of realtime, interactive voxel rendering in a "game" context. I posted one in the OP here. No commercial game AFAIK has used the tech yet, but that doesn't mean it isn't possible.