spatial data structures - university of southern...
TRANSCRIPT
![Page 1: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/1.jpg)
1
March 28, 2011Jernej Barbic
CSCI 480 Computer GraphicsLecture 17
Spatial Data Structures
Hierarchical Bounding VolumesRegular GridsOctreesBSP Trees[Ch. 10.12]
University of Southern Californiahttp://www-bcf.usc.edu/~jbarbic/cs480-s11/
![Page 2: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/2.jpg)
2
Ray Tracing Acceleration
• Faster intersections– Faster ray-object intersections
• Object bounding volume• Efficient intersectors
– Fewer ray-object intersections• Hierarchical bounding volumes (boxes, spheres)• Spatial data structures• Directional techniques
• Fewer rays– Adaptive tree-depth control– Stochastic sampling
• Generalized rays (beams, cones)
![Page 3: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/3.jpg)
3
Spatial Data Structures
• Data structures to store geometric information• Sample applications
– Collision detection– Location queries– Chemical simulations– Rendering
• Spatial data structures for ray tracing– Object-centric data structures (bounding volumes)– Space subdivision (grids, octrees, BSP trees)– Speed-up of 10x, 100x, or more
![Page 4: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/4.jpg)
4
Bounding Volumes
• Wrap complex objects in simple ones• Does ray intersect bounding box?
– No: does not intersect enclosed objects– Yes: calculate intersection with enclosed objects
• Common types:
Axis-alignedBounding
Box (AABB)
OrientedBoundingBox (OBB)
Sphere Convex Hull6-dop
![Page 5: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/5.jpg)
5
Selection of Bounding Volumes
• Effectiveness depends on:– Probability that ray hits bounding volume, but not
enclosed objects (tight fit is better)– Expense to calculate intersections with bounding
volume and enclosed objects• Amortize calculation of bounding volumes• Use heuristics
good
bad
![Page 6: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/6.jpg)
6
Hierarchical Bounding Volumes
• With simple bounding volumes, ray casting stillrequires O(n) intersection tests
• Idea: use tree data structure– Larger bounding volumes contain smaller ones etc.– Sometimes naturally available (e.g. human figure)– Sometimes difficult to compute
• Often reduces complexity to O(log(n))
![Page 7: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/7.jpg)
7
Ray Intersection Algorithm
• Recursively descend tree• If ray misses bounding volume, no intersection• If ray intersects bounding volume, recurse with
enclosed volumes and objects• Maintain near and far bounds to prune further• Overall effectiveness depends on model and
constructed hierarchy
![Page 8: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/8.jpg)
8
Spatial Subdivision
• Bounding volumes enclose objects, recursively• Alternatively, divide space (as opposed to objects)• For each segment of space, keep a list of
intersecting surfaces or objects• Basic techniques:
UniformSpatial Sub
Quadtree/Octree kd-tree BSP-tree
![Page 9: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/9.jpg)
9
Grids
• 3D array of cells (voxels) that tile space• Each cell points to all intersecting surfaces• Intersection
algorithmsteps from cellto cell
![Page 10: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/10.jpg)
10
Caching Intersection points
• Objects can span multiple cells• For A need to test intersection only once• For B need to cache intersection and check
next cell for any closer intersection withother objects
• If not, C could bemissed (yellow ray)
AB
C
![Page 11: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/11.jpg)
11
Assessment of Grids
• Poor choice when world is non-homogeneous• Size of grid
– Too small: too many surfaces per cell– Too large: too many empty cells to traverse– Can use algorithms like Bresenham’s
for efficient traversal• Non-uniform spatial subdivision more flexible
– Can adjust to objects that are present
![Page 12: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/12.jpg)
12
Outline
• Hierarchical Bounding Volumes• Regular Grids• Octrees• BSP Trees
![Page 13: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/13.jpg)
13
Quadtrees
• Generalization of binary trees in 2D– Node (cell) is a square– Recursively split into 4 equal sub-squares– Stop subdivision based on number of objects
• Ray intersection has to traverse quadtree• More difficult to step to next cell
![Page 14: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/14.jpg)
14
Octrees
• Generalization of quadtree in 3D• Each cell may be split into 8 equal sub-cells• Internal nodes store pointers to children• Leaf nodes store list of surfaces• Adapts well to non-homogeneous scenes
![Page 15: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/15.jpg)
15
Assessment for Ray Tracing
• Grids– Easy to implement– Require a lot of memory– Poor results for non-homogeneous scenes
• Octrees– Better on most scenes (more adaptive)
• Alternative: nested grids• Spatial subdivision expensive for animations• Hierarchical bounding volumes
– Natural for hierarchical objects– Better for dynamic scenes
![Page 16: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/16.jpg)
16
Other Spatial Subdivision Techniques
• Relax rules for quadtrees and octrees• k-dimensional tree (k-d tree)
– Split at arbitrary interior point– Split one dimension at a time
• Binary space partitioning tree (BSP tree)– In 2 dimensions, split with any line– In k dims. split with k-1 dimensional hyperplane– Particularly useful for painter’s algorithm– Can also be used for ray tracing
![Page 17: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/17.jpg)
17
Outline
• Hierarchical Bounding Volumes• Regular Grids• Octrees• BSP Trees
![Page 18: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/18.jpg)
18
BSP Trees
• Split space with any line (2D) or plane (3D)• Applications
– Painters algorithm for hidden surface removal– Ray casting
• Inherent spatial ordering given viewpoint– Left subtree: in front, right subtree: behind
• Problem: finding good space partitions– Proper ordering for any viewpoint– How to balance the tree
![Page 19: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/19.jpg)
19
Building a BSP Tree
• Use hidden surface removal as intuition• Using line 1 or line 2 as root is easy
Line 2Line 3
Line 1
Viewpoint
11
2
3
B A C D
a BSP treeusing 2 as root
A
B
D
C
3 2
the subdivisionof space it implies
![Page 20: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/20.jpg)
20
Splitting of surfaces
• Using line 3 as root requires splitting
Line 2a
Line 3
Line 1
Viewpoint
1
2a2b
Line 2b
3
![Page 21: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/21.jpg)
21
Building a Good Tree
• Naive partitioning of n polygons yields O(n3)polygons (in 3D)
• Algorithms with O(n2) increase exist– Try all, use polygon with fewest splits– Do not need to split exactly along polygon planes
• Should balance tree– More splits allow easier balancing– Rebalancing?
![Page 22: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/22.jpg)
22
Painter’s Algorithm with BSP Trees
• Building the tree– May need to split some polygons– Slow, but done only once
• Traverse back-to-front or front-to-back– Order is viewer-direction dependent– What is front and what is back of each line changes– Determine order on the fly
![Page 23: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/23.jpg)
23
Details of Painter’s Algorithm
• Each face has form Ax + By + Cz + D• Plug in coordinates and determine
– Positive: front side– Zero: on plane– Negative: back side
• Back-to-front: inorder traversal, farther child first• Front-to-back: inorder traversal, near child first• Do backface culling with same sign test• Clip against visible portion of space (portals)
![Page 24: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/24.jpg)
24
Clipping With Spatial Data Structures
• Accelerate clipping– Goal: accept or rejects whole sets of objects– Can use an spatial data structures
• Scene should be mostly fixed– Terrain fly-through– Gaming
Hierarchical bounding volumes Octrees
viewing frustrum
![Page 25: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/25.jpg)
25
Data Structure Demos
• BSP Tree constructionhttp://symbolcraft.com/graphics/bsp/index.html
• KD Tree constructionhttp://donar.umiacs.umd.edu/quadtree/points/kdtree.html
![Page 26: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/26.jpg)
26
Real-Time and Interactive Ray Tracing
• Interactive ray tracing via space subdivisionhttp://www.cs.utah.edu/~reinhard/egwr/
• State of the art in interactive ray tracinghttp://www.cs.utah.edu/~shirley/irt/
![Page 27: Spatial Data Structures - University of Southern Californiarun.usc.edu/cs480-s11/lec17-spatial-datastructures/17... · · 2016-06-26other objects •If not, C could be missed (yellow](https://reader035.vdocuments.net/reader035/viewer/2022070610/5af987f27f8b9a19548cc033/html5/thumbnails/27.jpg)
27
Summary
• Hierarchical Bounding Volumes• Regular Grids• Octrees• BSP Trees