I posted a curiosity thread on Vavoom forums a couple of days ago. Since then, I tried to learn a bit more about this technology, and found it was kind of a viable work area in 2000, bit since then only Build's creator has been actively trying to really dig this technology, and came up with an open source engine called Voxlap, that has support for gaming (even comes with a game demonstration and editors, but I can't check them out because appearently they need to be compiled o something).
Anyway, if you guys read the thread, you know Janis has stated that Voxels are indeed possible to be rendered by Vavoom, it seems Doom and Build's engine are easily compatible with it. But his question was: why would anyone want support for it?
Well, recent talks about mapping for SE brought to the surface the limitations provided by simple sector maps in terms of sheer size and details (complexity). If what I read is right, Voxels 3d maps were used SPECIALLY becuase they were simple enough to allow for incredibly large scenes to be rendered even when computer graphics were very primitive (which was impossible for polygons at the time).
I consider myself still completely ignorant about voxel mapping, so I'm not saying this will be the answer to our prayers, but this system does seems to have a few good qualities that make it a VERY viable contender for our TSP ("The Serpent Power" project) and even SE needs:
- First and foremost, IF Janis was right, this feature shouldn't be that hard to implement, meaning that we can actually USE voxel while most polygon terran generators are not compatible, or would probably bring issues Vavoom is not ready to support.
- Some tutorials suggest alpha images to generate heightmap information, which means voxels, depending on mapping possibilities, can generate complex landscapes that we need with methods MUCH less time-demanding as hand-draw sectors, sloping, etc. My rough guess is that we could actually use the original 2d map RambOrc created, add the tones to indicate heights, and convert it directly to 3d volmun!!!
- It arguably can render enormous amount of scenery without considerable cpu usage (not today's standard) - which means no more skyboxes tricks to emulate distant landscape - the landscape would actually be real.
- CAN be mixed successfully with polygon models (the normal 3D objects like Houses, decorations, monsters, trees, etc)
- Mixing with Sectors mechanics are yet to be explained or demonstrated (Build allowed it, don't know how it ould be in Vavoom, again, if at all).
- No support from 3d graphic card hardware. I have been unable to determine if today's machines are capable or not to render LARGE scenery with good detail without any problems (it seems so).
- The actualy maps look a little like blurred images in 3d (details and chaos are hard to come by it seems).
- Bilinear dithering is great to brake the "square" look, but that is an additional feature (meaning a Vavoom might or might not be able to do it, if voxels are implemented at all) = risk of having somewhat blocky landscape.
- Hard to implement textures on the "surfaces" created - many forms of renders doesn't seem to support anything other than pure colors (and shades) at all. Surfaces might look to plain, but it all omes down to what Janis would be able to actually implement.
Anyway, here are some good examples of voxels usage on modern games (circa 2000) mainly on Delta Force 2 and Outcast:
Voxel grass, volumn mixed with 3D models
HUGE landscape (don't kno if this is DF2 though)
Fog + Grass
So, the point here is: do you guys (specially RambOrc, as the leader) think Voxels are indeed a viable feature that we would indeed put to good use on the future project, thus making this a worthy request to put on Janis's appreciation?
My opinion is yes - Janis said he could do it, I think we could use, this has very good potential in my humbl?e opinion, I think this could REALLY work. I thought I should make a poll because I don't want to waste Janis' time and don't want to talk for the team.