Unlimited Detail, octree traversals

[Silent_Buddha]

A game like Final Fantasy XIII would be a good candidate then; almost all of the environments are static and non-interactive. One question, is it possible to combine a voxel based engine with a polygon one? I mean use voxel for the static parts of the world and polygons for the dynamic parts, again using Final XIII as and example where most of the backgrounds you can only see but not interact with.

Yes, I believe some of the unreleased ATI demos (Ruby Demo with the Robot, for example) have used Voxel + Polygon rendering. It was done by a company that specializes in such things (thus why ATI can't release it).

Regards,
SB

 

[Laa-Yosh]

...

Actually, in our case it's still more about dynamics. Almost everything we render as real 3D geometry has to animate and deform somehow - if something's static, we can just use matte paintings which are a LOT cheaper in every way then modeling-texturing-lighting-rendering. So there aren't really any cases where voxels would make sense in a dynamic 3D world.

Games may be an exception in that the camera is dynamic, but a lot of the environment can stay static in some cases. Unless, of course, you want to change the lighting...

 

[MfA]

Voxel models can be animated. Changing the lighting is not really more or less of a problem (which is to say it's always a big problem).

 

[Laa-Yosh]

I'm not aware of methods to replace skinning, blend shapes, muscle and cloth simulations etc. that are well-developed techniques on polygon based models...

As for the lighting, I've meant that most examples of voxel based content we've seen in tech demos and presentations have stored a single texture value that represented both the color and the amount of precalculated static lighting for that point of the object. In such an engine I don't see how the lighting could be changed.
I imagine it's perfectly possible to light voxel based geometry as well, but it'll mean a considerable amount of slowdown for the tech. Then again, rendering speed is the one thing we can count on to improve with time.

 

[MfA]

It's not really a question of replacing ... the same methods can apply, a bone can influence the position of a voxel as easily as the position of a vertex (of course the raycasting of voxels moved around in this way loses some of it's elegance).

 

[3dcgi]

Yeah, you have to wonder why movie VFX has not transitioned completely to voxels if they're the better solution for unlimited scene complexity.

My understanding is that for Pixar level movies shading dominates the render time so maybe there's not much to be gained from switching to voxels/points since the shading cost is still there. Besides as you've alluded to in another post the tools all work with polygons.

 

[Danalys]

I'd just go for as low geometry as you need to animate. Displacement maps on that but with each pixel representing a vertex. The limit for detail is going to be memory soon enough anyway.

 

[Grall]

It was done by a company that specializes in such things (thus why ATI can't release it).

That makes no sense at all... Why wouldn't they be able to release a work THEY commissioned - and I assume - paid for, regardless of what the other company specializes in?

 

[MfA]

If they paid for a graphical demo they don't necessarily get the distribution rights to an executable. That said, mr. Urbach always feels like a bit of a snake oil salesman.

Otoy is putting out some news again though.

 

[LightHeaven]

Isn't this almost exactly like what Ati has shown in that whole cinema 2.0 demos?

 

[brain_stew]

What I don't get is why this is spreading around the internet like wildfire once again, when the demos are the same ones they were showing 2 years ago? Investor drive perhaps? The fact that they appear to have made no progress at all in the last 2 years is rather worrying.

 

[DRS]

Quote:
Unlimited Detail is a software algorithm that gives unlimited geometry. When we say “unlimited geometry” we really do mean it. It really is Unlimited, Infinite, endless power, for 3D graphics.

I stopped reading there.

 

[Kittonwy]

Is this a technology that shows promise for this and next-gen consoles? It seems too good to be true, so what’s the catch.

http://www.youtube.com/watch?v=Q-ATtrImCx4&feature=player_embedded
http://unlimiteddetailtechnology.com/videos.html

Can I order some snake oil from these guys?

 

[Rolf N]

What I don't get is why this is spreading around the internet like wildfire once again, when the demos are the same ones they were showing 2 years ago? Investor drive perhaps? The fact that they appear to have made no progress at all in the last 2 years is rather worrying.

The average human being is not very smart, unfortunately.

 

[ShootMyMonkey]

My understanding is that for Pixar level movies shading dominates the render time so maybe there's not much to be gained from switching to voxels/points since the shading cost is still there.

Yeah, but shading is relatively easy to parallelize as well since a lot of what you need to know is set up beforehand. Well, Pixar has it a little harder since their pipeline can potentially modify/create geometry at shade time. Shading of voxels would theoretically be a completely deferred shading pipeline and you can know exactly which voxels are actually visible when you get around to shade time... so shading is pretty easy to parallelizable until you hit the wall on data throughput.

Though with a completely deferred pipeline, the cost will basically shift over to setup.

 

[DieH@rd]

They are back.
http://www.youtube.com/watch?v=UKUuUvDSXk4

 

[Tap In]

They are back.
http://www.youtube.com/watch?v=UKUuUvDSXk4

What the hell did I just watch?

 

[dragonelite]

What the hell did I just watch?

The same as last year...

 

[Deleted member 13524]

I wonder how many hats will be eaten if this actually comes through..

 

[fellix]

Impressive by its own, but still they are showing a completely static environment.