Realtime Shading with Filtered Importance Sampling

1
Real-time Shading with Filtered Importance
Sampling

• Jaroslav Krivánek
• Czech Technical University in Prague
• Mark Colbert
• University of Central Florida

2
Motivation

• material design interfaces
• rendering algorithm to back up the interface
• immediate feedback

3
Goal

• image-based lighting (environment maps)
• improves material perception Fleming et al.
  2003

images Fleming et al. 2003
natural light
point light
4
Goal

• arbitrary materials
• low to high gloss, different BRDF models

images by Pat Hanrahan
5
Goal

• dynamic materials, geometry, lighting
• no pre-computation
• production pipeline friendly
• minimal code base / single GPU shader
• real-time shadows (env. map)
• not necessarily

6
Desired Results
7
Related Work

• pre-filtered environment maps Kautz et al. 2000
  
• frequency-space rendering Ramamoorthi et al.
  2002 , Ng et al. 2004
• Efficient Reflectance and Visibility
  Approximations for Environment Map Rendering
  Green et al. 2007
• Efficient Rendering of Spatial Bi-directional
  Reflectance Distribution Functions McAllister
  et al. 2002

8
Overview

• Motivation
• Goal
• Related Work
• Shading Algorithm
• Theory
• Real-time Shadows
• Results
• Conclusion

9
BRDF Importance Sampling

• standard in MC ray tracing
• not used on the GPU

10
Deterministic Sampling

• aliasing

40 samples per pixel
11
Our Approach

• filtered importance sampling
• less filtering where samples are denser
• more filtering where they are sparser

12
Filtering

• MIP-maps
• level proportional to log of filter size
• independent of the BRDF

13
Filtered Importance Sampling
40 samples per pixel
14
Overview

• Motivation
• Goal
• Related Work
• Shading Algorithm
• Theory
• Real-time Shadows
• Results
• Conclusion

15
Underlying Theory

• why theory?
• identify approximations
• suggest improvements
• sampling filtering
• signal processing

16
Sampling and Reconstruction
sample
reconstruct
aliased original
Krivánek, ColbertReal-time shading with Filtered
Importance Sampling
16
17
Application to Importance Sampling

• problem non-uniform samples

18
Conceptual Procedure
19
Practice

• isotropic filter approximation

Krivánek, ColbertReal-time shading with Filtered
Importance Sampling
19
20
Approximations

• isotropic filter shape
• constant BRDF / PDF ratio across filter support
• tri-linear filtering (MIP-map)

21
Anisotropic Filtering Experiments

• anisotropic filter approximation

Krivánek, ColbertReal-time shading with Filtered
Importance Sampling
21
22
Anisotropic Filtering Experiments

• tex2Dgrad for anisotropic texture look-up
• worse image quality still dont know why

16x anisotropic filter
Krivánek, ColbertReal-time shading with Filtered
Importance Sampling
22
23
Overview

• Motivation
• Goal
• Related Work
• Shading Algorithm
• Theory
• Real-time Shadows
• Results
• Conclusion

24
Real-time Shadows

• environment importance sampling (bright light
  sources strongest shadows)

25
Real-time Shadows

• shadow map for each sample

26
Real-time Shadows

• convert to spherical harmonics at each texel

visibility function
27
Real-time Shadows

• spatial filtering

no filtering
after filtering
28
Real-time Shadows

• use for rendering
• diffuse
• SH coefficient dot product
• glossy
• attenuate each sample by the visibility

29
Overview

• Motivation
• Goal
• Related Work
• Shading Algorithm
• Theory
• Real-time Shadows
• Results
• Conclusion

Krivánek, ColbertReal-time shading with Filtered
Importance Sampling
29
30
FIS Results RMS
Environment Sampling
Filtered Sampling
Reference
n 3 5 Samples
n 17 100 Samples
31
FIS Results Complex Geometry
5 Samples
50 Samples
50 Samples
200 Samples
Reference
32
FIS Results BRDF Anisotropy

• limited anisotropy

ax 0.01 ax 0.08
ax 0.01 ax 0.29
ax 0.01 ax 0.01
33
Shadows Results
Reference (30,000 Samples)
Our Method (16 Samples)
Visual
Error
SH v. Ref
8 samples
16 samples
64 samples
34
Shadows Performance

• NVIDIA GeForce 8800 GTX, Intel Core2 Duo, 512x512

35
Shadows Performance

• NVIDIA GeForce 8800 GTX, Intel Core2 Duo, 512x512

polygon count
Krivánek, ColbertReal-time shading with Filtered
Importance Sampling
35
36

• Video

37
Conclusion

• glossy surface shading
• practical, relatively accurate, no
  pre-computation
• signal processing theory
• shadows
• fast but very approximate
• no pre-computation
• implementation details GPU Gems 3
• Code graphics.cs.ucf.edu/gpusampling/

38
Acknowledgements

• Dan Sýkora
• Petr Olák
• Czech Ministry of Education
• Center for Computer Graphics
• Aktion grant
• US National Science Foundation

39
Additional Slides
40
Numerical Integration as Signal Reconstruction

• integral DC term
• integration by sampling
• sample the function
• reconstruct the DC term
• insufficient sampling -gt aliasing
• alias may affect the DC term -gt error
• anti-aliasing pre-filtering

41
Anti-aliasing by Pre-filtering
band-limit
sample
reconstruct
42
Stochastic Sampling

• noise
• slower on the GPU

40 samples per pixel
Krivánek, ColbertReal-time shading with Filtered
Importance Sampling
42