Interactive Sampling and Rendering for Complex and Procedural Geometry

1
Interactive Sampling and Rendering for Complex
and Procedural Geometry

• Marc Stamminger, George Drettakis
• REVES/iMAGIS Sophia-Antipolis

2
motivation

• tree created by AMAP
• 150,000 triangles
• 8 fps(Linux PC with GeForce Quadro)

3
motivation

• rendered withpoints at 60 fps
• reduced quality
• 7 times faster

4
motivation

• level of detail
• 100 trees
• 270,000 points
• 20 fps

5
previous work

• 1985 Levoy/Whitted "The Use of Points as a
  Display Primitive"
• 1998 Grossman/Dally "Point Sample Rendering"
• 2000 Rusinkiewicz/Levoy "The Q-Splat"
• 2000 Pfister/Zwicker/van Baar/Gross "Surfels"

6
very recent work

• Wand/Fischer/Peter/Meyer/auf der Heide/Strasser
• "The Randomized z-Buffer Algorithm"

7
point rendering pipeline

• scene description
• vrml file
• mgf file
• procedural model

point set (3D-coordinates, normal, material)
screen
point rendering
point generation
8
point generation

• (orthographic) views
• filtered triangle mesh hierarchy
• random points

9
point rendering

• in software
• filtering
• texturing
• hole filling
• in hardware
• as points
• as polygonal disks

10
our approach

• fast / on the fly point generation for
• procedural objects
• terrains
• complex dynamic objects
• point rendering with OpenGLs GL_POINT
• very fast (up to 10 million points per second)
• OpenGL does lighting

11
results

• points are well suited for
• proceduralgeometry

12
results

• points are well suited for
• proceduralgeometry
• terrains

13
results

• points are well suited for
• proceduralgeometry
• terrains
• complexgeometry

14
results

• points are well suited for
• proceduralgeometry
• terrains
• complexgeometry
• combinations

15
complex polygonal geometry

• generate list of randomly distributed samples
• for every frame compute n, render the first n

10,000
1,000
100,000
16
complex polygonal geometry

• easy speed / quality trade off
• frame rate control

10,000
1,000
100,000
17
modified complex geometry

• simple modifications on the fly

30 fps
18
displaced geometry
25,000 points
25,000 points
19
displaced geometry
25,000 points
100,000 points
20
adaptive sampling
21
undersampling factor
lt 1
gt 1
22
undersampling factor
23
video sqrt(5) sampling
24
adaptive point generation
25
sqrt(5) sampling
(2/5,1/5)
26
sqrt(5) sampling
27
sqrt(5) sampling
28
sqrt(5) sampling
29
sqrt(5) sampling
30
sqrt(5) sampling
31
sqrt(5) sampling

• rotated, nested grids
• grid distance decreases by 1/sqrt(5)
• rotation angle ? 27o
• special attention to boundaries

32
procedural modifiers
33
terrains
34
terrains
35
terrain parameterization

• parameterize by (d,u)

terrain
d
u
screen
36
terrain parameterization
looking straight ahead
looking up
looking down
37
terrain algorithm

• sqrt(5) sampling scheme
• undersampling factor
• parameterizationdistortions
• perspectivedistortions
• displacement

38
terrain occlusion culling

• elevation directionin image spacealong v
• simplifies occlusionculling

elevation
39
terrain occlusion culling
occlusion culling, regular sampling
occlusion culling, with adaptive sampling
40
video results
41
conclusion

• points are very powerful, when details
  becomesmaller than a pixel
• simple and efficient level of detail
• simple manipulation
• easily parallelizable
• big potential for further improvements

42
link

• more at
• http//www-sop.inria.fr/reves

43
acknowledgements

• thanks to the European Union for funding me
• with a Marie-Curie fellowship

44
(No Transcript)