Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing...
Transcript of Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing...
![Page 1: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/1.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
![Page 2: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/2.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
A
D
C B
X
Y1
X
Y1 Y2
A B C D
tmin
tmax
Y2
![Page 3: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/3.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
A
D
C B
X
Y1
X
Y1 Y2
A B C D
tmin
tmax
Y2
Y2
![Page 4: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/4.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
A
D
C B
X
Y1
X
Y1 Y2
A B C D
tmin
tmax
Y2
![Page 5: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/5.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Y
? ?
? ? ? ?
tmin
tmax
Y
A large tree structure change. A totally new tree!
![Page 6: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/6.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Solution 1: Rebuild kd-tree each frame Rebuild kd-tree in a lazy manner, approximate SAH (Surface
Area Heuristics) [Hunt et al. 06] Can just move objects bounding boxes around and
transform rays (for hierarchical movement) [Wald et al. 03] Motion decomposition, fuzzy kd-trees [Günther et al. 06]
Solution 2: use different hierarchical structure
![Page 7: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/7.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Bounding volume hierarchies (BVHs) [Wald et al. 06b, Boulos et al. 06, Lauterbach et al. 06]
Grids [Wald et al. 06a]
![Page 8: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/8.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Ray Tracing Dynamic Scenes Using BVHs [Lauterbach et al. 06]
Dinesh Manocha, Christian Lauterbach University of North Carolina at Chapel Hill
![Page 9: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/9.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Tree of bounding volumes (sphere, AABB, OBB, k-DOP, spherical shells, etc.)
Each bounding volume encloses “nearby” primitives
Parent node primitives are union of children node primitives
![Page 10: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/10.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Spatial partitioning: space is subdivided into disjoint regions (e.g. grid,
kd-tree, octree, ...) Object hierarchy:
groups or clusters of objects/primitives are subdivided (BVH, s-kd-tree)
![Page 11: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/11.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Implications for ray tracing Spatial partitioning: Objects referenced in multiple
nodes (overlap in object space) BVH Hierarchies: Nodes can overlap each other
(overlap in 3D space) Spatial partitioning allows easier front-to
-back ordering
![Page 12: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/12.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Widely used for intersection computations Ray tracing Visibility culling: view frustum and occlusion culling Collision and proximity computations Other applications
![Page 13: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/13.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Pretty simple: Start from root If ray intersects AABB, try all children, too:
is inner node: recurse on both children is leaf node: intersect with primitive(s)
Naïve implementation far slower than kd-tree!
![Page 14: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/14.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Intersection test more costly Up to 6 ray-plane intersections for AABB (slabs
test) Just 1 for kd-tree
No front-to-back ordering Cannot stop after finding first hit
Nodes take more space 32 bytes vs. 8 bytes
![Page 15: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/15.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
AABBs can provide tighter fit automatically No empty leafs, tree does not need to be as deep Primitives only referenced once ⇒ less nodes in hierarchy
#nodes known in advance (2n-1) (if 1 primitive/leaf)
![Page 16: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/16.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
AABBs can provide tighter fit automatically No empty leafs, tree does not need to be as deep Primitives only referenced once ⇒ less nodes in hierarchy
#nodes known in advance (2n-1) (if 1 primitive/leaf)
Can be updated easily!
![Page 17: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/17.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
What does updating mean? Underlying geometry changes Update will ensure correctness of hierarchy
without rebuilding it Should be faster than rebuild
![Page 18: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/18.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Post-order traversal of BVH Update children's AABB, then update own At leaf level, update from primitives Also update additional information such as axis
O(n) time Usually a few ms for small scenes May become too long for large models!
![Page 19: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/19.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Quality of BVH may decrease over animation Update does not change tree topology Rebuild may be necessary How to detect?
In worst-case scene: Performance dropping an order of magnitude over
20 animation frames Not as bad for normal scenes, though
![Page 20: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/20.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Use heuristic to detect degradation Assume performance lower when BVHs
contain lots of empty space:
![Page 21: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/21.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
How to measure quality? Use ratio of surface area parent to children
SA(parent) / ( SA(child1) + SA(child2) ) Save on rebuild for each node (4 bytes/node)
On each update: compare to initial value Sum up differences and normalize If above threshold: initiate rebuild
~30-40% work well in practice
![Page 22: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/22.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Video
![Page 23: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/23.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Ray Tracing Animated Scenes using Coherent Grid Traversal
[Wald et al. 06a]
I Wald, T Ize, A Kensler, A Knoll, S Parker SCI Institute, University of Utah
![Page 24: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/24.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
A new traversal techniques for uniform grids … that makes packet/frustum traversal
compatible with grids … thus achieves performance competitive with
fastest kd-trees … and which allows for per-frame rebuilds
(dynamic scenes)
![Page 25: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/25.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Since 70’ies: Lots of different RT data structures
![Page 26: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/26.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH
Since 70’ies: Lots of different RT data structures
![Page 27: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/27.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH
Since 70’ies: Lots of different RT data structures
![Page 28: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/28.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH
Since 70’ies: Lots of different RT data structures
![Page 29: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/29.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Since 70’ies: Lots of different RT data structures
![Page 30: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/30.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Since 70’ies: Lots of different RT data structures
![Page 31: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/31.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Since 70’ies: Lots of different RT data structures
![Page 32: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/32.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Since 70’ies: Lots of different RT data structures
![Page 33: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/33.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Grid
Since 70’ies: Lots of different RT data structures
![Page 34: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/34.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Grid
Since 70’ies: Lots of different RT data structures
![Page 35: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/35.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Grid
Since 70’ies: Lots of different RT data structures
![Page 36: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/36.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Grid
Since 70’ies: Lots of different RT data structures
![Page 37: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/37.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Grid Kd-tree
Since 70’ies: Lots of different RT data structures
![Page 38: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/38.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Grid Kd-tree
Since 70’ies: Lots of different RT data structures
![Page 39: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/39.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Grid Kd-tree
Since 70’ies: Lots of different RT data structures
![Page 40: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/40.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Grid Kd-tree
Since 70’ies: Lots of different RT data structures
![Page 41: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/41.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
BVH Octree
Grid Kd-tree Of all these, grid is only that is not hierarchical !
Since 70’ies: Lots of different RT data structures
![Page 42: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/42.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid is not hierarchical… Much simpler to build (similar to 3D
-rasterization, very fast) Build-times in the paper: 2.2M “Soda Hall” in 110 ms
Ideally suited for handling dynamic scenes Full rebuild every frame, no restrictions at all !
![Page 43: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/43.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
All of the recent advancements of RT are for kd-trees ! Pre-2000: Tie between grids and kd-trees… [Wald ’01]: New concept “coherent ray tracing” (for kd
-tree) Trace “packets” of coherent rays 10x faster than single rays
[Woop ’05]: First RT hardware prototype RPU (for kd-tree) [Reshetov ’05]: New concept “multilevel ray tracing” (kd
-tree) Trace packets using bounding frusta another 10x faster than
CRT !
But: (good) kd-trees are (too) costly to build…
![Page 44: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/44.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
SIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT:
~100fps) If ray tracing is to ever replace rasterization, it
must support dynamic scenes (games…) But: All our fast RT algos are for kd-trees… … and kd-trees can’t do dynamic scenes …
![Page 45: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/45.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
SIGGRAPH ‘05: Dynamic Scenes huge problem Since then, lots of research
Lazy kd-tree construction (Razor [Stoll, Mark ‘06]) Fast BVH and kd-tree construction (yet unpublished) Motion decomposition [Günther et al. ‘06] Dynamic BVHs [Wald et al. ‘06, Lauterbach et al. ’06] Hybrid BVH/kd-trees [Woop ‘06, Havran ‘06, Wachter ‘06,
…] Coherent Grid Traversal [Wald et al. ’06]
![Page 46: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/46.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
2005: Grid too slow to traverse (vs kd-tree)… Fact: Fast RT needs “packets” & “frusta”
concepts Traverse multiple packets over same node of DS
Rather simple for hierarchical data structures… Test both children in turn for overlap w/ packet If child overlaps: traverse it, else: skip it.
(it’s as simple as that)
… but not for grids
![Page 47: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/47.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Packets & grids: “Non-trivial task” In which order to test the nodes ? ABCD or
ABDC ? What to do when packet diverges?
3DDDA etc break in that case… Split diverging packet ?
Quickly degenerates to single-ray traversal…
Fix by re-merging packets ? Non-trivial & costly …
![Page 48: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/48.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
First: Transform all rays into “canonical grid space” i.e., [0,0,0]-[Nx,Ny,Nz]
![Page 49: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/49.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays”
![Page 50: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/50.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays”
![Page 51: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/51.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell”
Pick “major traversal axis” (e.g., max component of 1st ray)
![Page 52: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/52.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell”
![Page 53: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/53.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell”
![Page 54: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/54.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell”
![Page 55: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/55.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell”
![Page 56: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/56.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell” For each slice, compute frustum/slice overlap
![Page 57: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/57.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell” For each slice, compute frustum/slice overlap
![Page 58: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/58.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell” For each slice, compute frustum/slice overlap
Float-to-int gives overlapped cell IDs
![Page 59: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/59.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell” For each slice, compute frustum/slice overlap
Float-to-int gives overlapped cell IDs Intersect all cells in given slice
![Page 60: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/60.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell” For each slice, compute frustum/slice overlap
Float-to-int gives overlapped cell IDs Intersect all cells in given slice
Loop: incrementally compute next slice’s overlap box 4 additions…
![Page 61: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/61.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Idea: Consider only frustum, not “set of rays” Traverse “slice by slice” instead of “cell to cell” For each slice, compute frustum/slice overlap
Float-to-int gives overlapped cell IDs Intersect all cells in given slice
Loop: incrementally compute next slice’s overlap box 4 additions…
![Page 62: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/62.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Expensive setup phase Transform rays to canonical grid coordinate system Determine major march direction (simple) Compute min/max bounding planes (slopes and offsets) Compute first and last slice to be traversed (full frustum clip)
But: Very simple traversal step Overlap box update: 4 float additions (1 SIMD instruction) Get cell IDs: 4 float-to-int truncations (SIMD…) Loop over overlapped cells (avg: 1.5-2 cells per slice)
![Page 63: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/63.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid usually less efficient than kd-tree
![Page 64: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/64.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid usually less efficient than kd-tree Cannot adapt to geometry as well more
intersections
![Page 65: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/65.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid usually less efficient than kd-tree Cannot adapt to geometry as well more
intersections Tris straddle many cells re-intersection
![Page 66: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/66.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid usually less efficient than kd-tree Cannot adapt to geometry as well more
intersections Tris straddle many cells re-intersection
First sight: Frustum makes it worse…
![Page 67: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/67.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid usually less efficient than kd-tree Cannot adapt to geometry as well more
intersections Tris straddle many cells re-intersection
First sight: Frustum makes it worse… Rays isec tris outside “their” cells
![Page 68: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/68.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid usually less efficient than kd-tree Cannot adapt to geometry as well more
intersections Tris straddle many cells re-intersection
First sight: Frustum makes it worse… Rays isec tris outside “their” cells Re-isec aggravated by width of frustum
![Page 69: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/69.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid usually less efficient than kd-tree First sight: Frustum makes it worse… But: Two easy fixes
![Page 70: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/70.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid usually less efficient than kd-tree First sight: Frustum makes it worse… But: Two easy fixes
Bad culling SIMD Frustum culling in Packet/Tri Isec [Dmitriev et al.] Outside frustum cull!
![Page 71: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/71.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Grid usually less efficient than kd-tree First sight: Frustum makes it worse… But: Two easy fixes
Bad culling SIMD Frustum culling in Packet/Tri Isec [Dmitriev et al.]
Re-intersection: Mailboxing [Haines]
Mailbox detects re-intersection
![Page 72: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/72.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Surprise: Mailboxing & Frustum culling very effective Both standard techniques, both limited success for
kd-trees Grid & Frustum: Exactly counter weak points of
CGT … “Hand”
Grid w/o FC & MB : 14 M ray-tri isecs Grid with FC & MB: .9 M ray-tri isecs (14x less) Kd-tree : .85M ray-tri isecs (5% less than grid)
And: cost indep of #rays very cheap (amortize)
![Page 73: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/73.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
![Page 74: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/74.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Comparison to single-ray grid Fast single-ray traverser, macrocell if
advantageous, … Speedup 6.5x to 20.9x, usually ~10x
Comparison to kd-tree To OpenRT: 2x-8x faster (2M Soda Hall: 4.5x) To MLRT: ~3x slower (but much less optimized) Tests performed on “kd-tree friendly” models
![Page 75: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/75.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Build time: Usually affordable even on single CPU…
Traversal results (1024^2, dual 3.2 GHz Xeon PC) X/Y: X=raycast only; Y=raytrace+shade+texture+shadows
“Hand” 16K triangles 34.5/15.3 fps
“Poser” 78K triangles 15.8/7.8 fps
“Fairy” 174K triangles
3.4/1.2 fps
“Marbles” 8.8K triangles 57.1/26.2 fps
“Toys” 11K triangles 29.3/10.2 fps
![Page 76: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/76.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Comparison to state-of-the-art BVH or kd-tree Somewhat harder to code and “get right” than, e.g., BVH Usually somewhat slower (~1.5x-3x) More susceptible to incoherence & teapot-in-stadium cases
Pure frustum tech.: Visits all cells in frustum even if not touched by any ray!
BUT: It works at all ! (Who’d have thought 12m ago ?) ~10x faster than single-ray grid Benefits better from additional coherence (4x AA at 2x cost) “Maybe” better suited for regular data or special HW (Cell,
GPUs) Most flexible wrt dynamic no limitation at all
![Page 77: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/77.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
Have developed a new technique that Makes grid compatible with packets &
frusta Is competitive with BVHs and kd-trees Most general in handling dynamic scenes
![Page 78: Acceleration Structure for Animated ScenesSIGGRAPH ‘05: Dynamic Scenes huge problem Ray tracing has become very fast (MLRT: ~100fps) If ray tracing is to ever replace rasterization,](https://reader033.fdocuments.net/reader033/viewer/2022060921/60ac90ec61827f42752fab59/html5/thumbnails/78.jpg)
Acceleration Structure for Animated Scenes
Copyright © 2010 by Yong Cao
[Boulos et al. 06]: Solomon Boulos, Dave Edwards, J Dylan Lacewell, Joe Kniss, Jan Kautz, Peter Shirley, and Ingo Wald. Interactive Distribution Ray Tracing. Technical Report, SCI Institute, University of Utah, No UUSCI-2006-022, 2006.
[Günther et al. 06]:Johannes Günther, Heiko Friedrich, Ingo Wald, Hans-Peter Seidel, and Philipp Slusallek. Ray tracing animated scenes using motion decomposition. Computer Graphics Forum, 25(3), September 2006 (to appear)
[Havran et al. 06]: Vlastimil Havran, Robert Herzog, and Hans-Peter Seidel. On Fast Construction of Spatial Hierarchies for Ray Tracing. Submitted to RT’06, 2006.
[Lauterbach et al. 06]: Christian Lauterbach, Sung-Eui Yoon, David Tuft, Dinesh Manocha. RT-DEFORM: Interactive Ray Tracing of Dynamic Scenes using BVHs. Technical Report TR06-10, University of North Carolina at Chapel Hill, 2006.
[Mahovsky and Wyvill 04]: Jeffrey Mahovsky, Brian Wyvill. Fast Ray-axis Aligned Bounding Box Overlap Tests with Plücker Coordinates. Journal of Graphics Tools, 9(1):35-46, 2004
[Reshetov et al. 05]: Alexander Reshetov, Alexei Soupikov, and Jim Hurley. Multi-level ray tracing algorithm. ACM Trans. Graph., 24(3):1176–1185, 2005.
[Rubin and Whitted 80]: Steven M. Rubin and Turner Whitted. A 3-dimensional representation for fast rendering of complex scenes. Computer Graphics, 14(3):110–116, July 1980.
[Smits98]: Brian Smits. Efficiency issues for ray tracing. Journal of Graphics Tools: JGT, 3(2):1–14, 1998. [Wächter and Keller 06]: Carsten Wächter and Andreas Keller. Instant Ray Tracing: The Bounding Interval Hierarchy. Rendering Techniques 2006: Eurographics Symposium on Rendering, 2006.
[Wald et al. 03]: Ingo Wald, Carsten Benthin, and Philipp Slusallek. Distributed Interactive Ray Tracing of Dynamic Scenes. In Proceedings of the IEEE Symposium on Parallel and Large-Data Visualization and Graphics (PVG), 2003.
[Wald et al. 06a]:Ingo Wald, Thiago Ize, Andrew Kensler, Aaron Knoll, and Steven Parker. Ray Tracing Animated Scenes using Coherent Grid Traversal. In ACM Transaction on Graphics (Proc. SIGGRAPH 2006).
[Wald et al. 06b]: Ingo Wald, Solomon Boulos, and Peter Shirley. Ray Tracing Deformable Scenes using Dynamic Bounding Volume Hierarchies. Technical Report, SCI Institute, University of Utah, No UUSCI-2005-014 (conditionally accepted at ACM Transactions on Graphics), 2006.
[Woop et al. 06]: Sven Woop, Gerd Marmitt, and Philipp Slusallek. B-KD Trees for Hardware Accelerated Ray Tracing of Dynamic Scenes. In Proceedings of Graphics Hardware (to appear), 2006.
[Hunt et al. 06]: Warren Hunt, William R. Mark and Gordon Stoll, Fast kd-tree Construction with an Adaptive Error-Bounded Heuristic 2006 IEEE Symposium on Interactive Ray Tracing.