Forum archive
Voxel Usage in Other Games
- I get the impression that some of the forum participants may not be aware that there have been a number of commercial games that have employed voxels in some fashion:
Infogrames Outcast
Westwood Studios Blade Runner
Novalogic Comanche (and others)
Reviewers were rather brutal in their comments concerning the use of this technology, and it is unfortunate that proponents will likely have to struggle to overcome such prejudice in order to gain acceptance.
If you just can't get enough of that voxel goodnes or you want to get a historical perspective on voxel usage, lay your hands on these games or their demos.
Good luck to anyone who pursues these types of alternative technologies. - I think ken probably owns all three of them, plus several more!
And for those people who don't appreciate the world of voxels, All I can say is that I feel sorry for you.
I think voxels look sexy. :D Voxels
Novalogic used to really push voxels. I remember that almost everything they released contained voxels in some form or other at one point. Delta Force was another prime example.Voxel Usage in Other Games
I found two more games (full downloads on the Underdogs):
CyberRace (http://www.the-underdogs.org/game.php?name=CyberRace)
1993 game by a company called Cyberdreams.
Strike Base(http://www.the-underdogs.org/game.php?id=3992)
1996 game by a company called Max Design. Download has no movies, sound, or music so I'm guessing it's really boring. :wink:
Can't tell the quality though since HotU uses small pictures.- Those Games only use Heightmap Voxel-Engines... that isn't really compareable to Voxlaps fully deformable Landscapes.
Voxlap Heightmaps aren't spectacular or very useful today, cause Heightmap Landscapes can be done with polygons in higher visual quality.
The real strength of Voxels are full deformable objects and landscapes, but they haven't been used in games yet (as far as i know) cause those Engines were too slow. But Ken's Voxlap Engine is a great step into the right direction... into a full deformable future. :)
I can't wait to get my hands on the engine to fiddle around with it. :D no
Well, in my opinion it's a total illusion that voxels are "better" than polygons.
I like them better "just because"... (i also like old school pixelgrafix better than vector illustrations;-))
But from a technical perspective, voxels are NOT better to create deformable landscape. You can achieve the same effect with polygonal engines, it just takes a bit of calculation when you blast a hole in some wall.Voxel Usage in Other Games
But from a technical perspective, voxels are NOT better to create deformable landscape. You can achieve the same effect with polygonal engines, it just takes a bit of calculation when you blast a hole in some wall.
I don't get you, thats exactly why voxels are in the technically sense better suited for destructable objects; you can deform objects *way* easier with voxels. Doing this with polygonal objects require more efford. Also with voxels you can describe solid objects unlike a polygonal mesh.
But on the other hand voxels are in technical sense less suited for big scene's (and thus games) because they take up insane amount of memory for your average (hires) model. They're just 3d bitmaps, because everything scales in 3d, you'd need an insane amount of voxels to prevent jaginess / total blurring when the object fills up a big part from the screen. (try to scale up a 320x240 jpeg 4 times to fill your 1280x960 screen)
Because of this high memory requirement for voxel data, I dont think we'll ever be seeing voxels be used for all geometry. I think for certain solid objects voxel data could be used which then be converted to a polygon mesh for rendering (is possible with algorithms like marching cubes), which also already has been used in a couple of games, for example worms3d.
Maybe when realtime raytracing is a possibility (www.openrt.de), volume data could be rendered using raytracing... just for each ray that hits a voxel, a pixel is drawn i think :)I guess it would be very elegant to mix implicit/polygonal/voxel models that way...But the insane memory requirement will always remain for non-generated voxel data :(- I hardly doubt voxels are good for something today. They are slow, and insanely memory hungry. I believe that processing thousands of voxel is rather brute force and eats lot of power, so you dont have it for oher things. Maps must be small. I believe that polygons will last, only in the future they will be generated on the fly from parametric surfaces feeded to the card. Polygons are more "very close" friendly too.
EDIT: Plus I can hardly imagine what for will be voxels more effective . You are unable to do skeletal animation or morphing with them. Only thing they might be used for is blowing holes into the wall but I think that in realilty CSG substraction might be better.
I once believed that PS HW voxels could be used to generate fine texture detail, but this was achieved by paralax displacement mapping (maybe that technique is voxel like :). - A pure Voxel Engine like Voxlap isn't a good choice, thats right. it would be perfect to use voxels as data storage for landscape deformation data, convert this data into a polygonal mesh for rendering and use pure polygon models for objects. very much like in worms3d but with a much higher voxel resolution.
To get a good performance a good LOD algorithm for the polygonal landscape mesh and an advanced compression method for the voxel data are essential. Raytracing through the voxel data wouldn't be neccesary in this method so that we are able to use a more complex and slower compression algorithm for the voxel data than in voxlap, cause the voxel data is only needed when we are rebuilding the polygonal mesh.
To speed up the mesh-rebuilding, the lanscape should be splitted into smaller pieces with a size of 128x128x128 or something like that. by doing so you'll only have to rebuild this part of the polygonal landscape mesh when the voxel data gets modified. additionally you could generate lower detail versions of those meshes to be used in the more distant areas of the landscape to reduce the polygon count. this would be a simple and quite effective way to handle the LOD of the landscape.