Why are voxels more efficient than polygons?

Started by
12 comments, last by johnchapman 8 years, 6 months ago

I recently watched this video, and while I found the concept interesting I'm still really not clear on what makes voxels so much more efficient than polygons. The artistic implications are pretty neat (voxels are undoubtedly a more flexible modeling medium than polygons. I'm not sure if they're flexible in a way games could actually make much use of, but that's another question), but how is describing a model as a set of cubes more efficient than describing it as a set of triangles?

The general consensus seems to be that you can put a lot more voxels onto the screen than polygons (though I assume this is greatly offset by the inherent inaccuracy of voxels), but I haven't found an explanation for why. If anyone could link me to a good video/article on the technical reason for this, I would appreciate it.

Even if they do truly offer a much greater level of geometric detail than polygons, from my own (admittedly limited) experience with graphics programming I've found post-processing to be a much greater bottleneck than rasterization, and I don't see how voxels would do much to change that.

If voxels (or a hybrid solution) are really the way of the future, why are polygonal engines still so dominant? Are there any major drawbacks that aren't being addressed? I'm genuinely curious.

Advertisement
I have no idea what that video is on about... It seems as if every single statement he makes is just pulled out of his arse?

If voxels (or a hybrid solution) are really the way of the future, why are polygonal engines still so dominant?

Engine developers will pursue any solution that is efficient and scalable, or solves the specific needs of their game.
GPU polygon rasterization is pretty damn efficient, so it tends to get used a lot. Lots of games do use "voxels", but then actually generate polygon data from their voxel data for rasterization! So it's not strictly a one-or-the-other deal.
Alternative rendering methods such as ray-tracing are starting to become more popular. UE4 has some experimental sparse-voxel-octree ray-casting tech, and signed-distance-field sphere-tracing tech. SVO ray-casting is a "voxel" technique, but SDF sphere-tracing is neither polygons nor voxels (nor subdivisions) - it's an implicit surface biggrin.png
However, neither of these techniques are used for primary rays, as rasterization is more efficient for that; they're used for reflection and shadow rays.

from my own (admittedly limited) experience with graphics programming I've found post-processing to be a much greater bottleneck than rasterization

Yep, that too. With deferred shading, all your expensive lighting/shading calculations happen in post too...

It reminds me of those "infinite detail" demos from years back.

I was pretty skeptical about some of the claims he made, like "reduce game development cost by up to 90%!"; yeah I don't think so. But either way, I see voxel rendering pop up from time to time in various places and I'm still curious about why they haven't caught on, if they're really as good as people claim (possibly because they aren't, I'm not sure).

It seems as if every single statement he makes is just pulled out of his arse?

It reminds me of those "infinite detail" demos from years back.

Yep, it looks like it's Unlimited Detail repackaged to lure investors.

Here's an old thread about it:
http://www.gamedev.net/topic/599083-unlimited-detail-technology/

The concept is that Instancing data-efficient voxels lets you render a lot of them.

The same is true for polygons, really.

Yep, it looks like it's Unlimited Detail repackaged to lure investors.

They already rebranded as Euclideon, and got a lot of investment.
Thankfully, their investors convinced them to actually market it at an industry where a massive point-cloud viewer is actually a useful product (GIS), and they're actually selling it to people there now.

And you know this isn't one of their videos because it doesn't have terrible bruce dell narration(tm).

I'm surprised that they don't realize that claims of reducing costs by 90%, having unlimited detail, etc., are just going to sound like huge red flags to people.

I'm surprised that they don't realize that claims of reducing costs by 90%, having unlimited detail, etc., are just going to sound like huge red flags to people.

Honestly, the only thing that'd convince me of it's worth is many different smooth voxel animations in realtime. I'm sure it's not impossible, but no one's come up with a really solid solution that I've seen before. I think atomage is the closest so far.

Freaky! Still, it looks fairly procedural. What about animating a dragon?

This topic is closed to new replies.

Advertisement