Download presentation
Presentation is loading. Please wait.
Published byErica Shaw Modified over 9 years ago
1
All-Frequency Shadows Using Non-linear Wavelet Lighting Approximation Ren Ng Stanford Ravi Ramamoorthi Columbia Pat Hanrahan Stanford
2
From Frank Gehry Architecture, Ragheb ed. 2001 Lighting Design
3
From Frank Gehry Architecture, Ragheb ed. 2001 Lighting Design
4
Existing Fast Shadow Techniques We know how to render very hard and very soft shadows Sloan, Kautz, Snyder 2002 Shadows from smooth lighting (precomputed radiance transfer) Sen, Cammarano, Hanrahan, 2003 Shadows from point-lights (shadow maps, volumes)
5
All-Frequency Lighting Teapot in Grace Cathedral
6
Soft Lighting Teapot in Grace Cathedral
7
All-Frequency Lighting Teapot in Grace Cathedral
8
Soft Lighting Teapot in Grace Cathedral
9
All-Frequency Lighting Teapot in Grace Cathedral
10
Overview Matrix formulation of relighting n Pre-computing the light-transport matrix n Compressing the matrix multiplication Results n Demonstration n Error analysis (Comparison with PRT) Future directions n Limitations n Areas for research
11
Overview Matrix formulation of relighting n Pre-computing the light-transport matrix n Compressing the matrix multiplication Results n Demonstration n Error analysis (Comparison with PRT) Future directions n Limitations n Areas for research
12
Relighting as Matrix-Vector Multiply
13
Input Lighting (Cubemap Vector) Output Image (Pixel Vector) Transport Matrix
14
Ray-Tracing Matrix Columns
16
Light-Transport Matrix Rows
19
Rasterizing Matrix Rows Pre-computing rows n Rasterize visibility hemicubes with graphics hardware n Read back pixels and weight by reflection function
20
Overview Matrix formulation of relighting n Pre-computing the light-transport matrix n Compressing the matrix multiplication Results n Demonstration n Error analysis (Comparison with PRT) Future directions n Limitations n Areas for research
21
Overview Matrix formulation of relighting n Pre-computing the light-transport matrix n Compressing the matrix multiplication Results n Demonstration n Error analysis (Comparison with PRT) Future directions n Limitations n Areas for research
22
Matrix Multiplication is Enormous Dimension n 512 x 512 pixel images n 6 x 64 x 64 cubemap environments Full matrix-vector multiplication is intractable n On the order of 10 10 operations per frame
23
Sparse Matrix-Vector Multiplication Choose data representations with mostly zeroes Vector: Use non-linear wavelet approximation on lighting Matrix: Wavelet-encode transport rows
24
Non-linear Wavelet Light Approximation Wavelet Transform
25
Non-linear Wavelet Light Approximation Non-linear Approximation Retain 0.1% – 1% terms
26
Why Non-linear Approximation? Linear n Use a fixed set of approximating functions n Precomputed radiance transfer uses 25 - 100 of the lowest frequency spherical harmonics Non-linear n Use a dynamic set of approximating functions (depends on each frame’s lighting) n In our case: choose 10’s - 100’s from a basis of 24,576 wavelets Idea: Compress lighting by considering input data
27
Why Wavelets? Wavelets provide dual space / frequency locality n Large wavelets capture low frequency, area lighting n Small wavelets capture high frequency, compact features In contrast n Spherical harmonics n Perform poorly on compact lights n Pixel basis n Perform poorly on large area lights
28
Choosing Non-linear Coefficients Three methods of prioritizing 1. Magnitude of wavelet coefficient n Optimal for approximating lighting 2. Transport-weighted n Biases lights that generate bright images 3. Area-weighted n Biases large lights
29
Sparse Matrix-Vector Multiplication Choose data representations with mostly zeroes Vector: Use non-linear wavelet approximation on lighting Matrix: Wavelet-encode transport rows Choose data representations with mostly zeroes Vector: Use non-linear wavelet approximation on lighting Matrix: Wavelet-encode transport rows
30
Matrix Row Wavelet Encoding
31
Extract Row
32
Matrix Row Wavelet Encoding Wavelet Transform
33
Matrix Row Wavelet Encoding Wavelet Transform
34
Matrix Row Wavelet Encoding Wavelet Transform
35
Matrix Row Wavelet Encoding Wavelet Transform
36
Matrix Row Wavelet Encoding Store Back in Matrix 0 ’ ’ 0 0
37
Matrix Row Wavelet Encoding 0 ’ ’ 0 0 Only 3% – 30% are non-zero
38
Total Compression Lighting vector compression n Highly lossy n Compress to 0.1% – 1% Matrix compression n Essentially lossless encoding n Represent with 3% – 30% non-zero terms Total compression in sparse matrix-vector multiply n 3 – 4 orders of magnitude less work than full multiplication
39
Overall Relighting Algorithm Pre-compute (per scene) n Compute matrix in pixel basis n Wavelet transform rows n Quantize, store Interactive Relighting (each frame) n Wavelet transform lighting n Prioritize and retain N wavelet coefficients n Perform sparse-matrix vector multiplication
40
Overview Matrix formulation of relighting n Pre-computing the light-transport matrix n Compressing the matrix multiplication Results n Demonstration n Error analysis (Comparison with PRT) Future directions n Limitations n Areas for research
41
Overview Matrix formulation of relighting n Pre-computing the light-transport matrix n Compressing the matrix multiplication Results n Demonstration n Error analysis (Comparison with PRT) Future directions n Limitations n Areas for research
42
Demo
43
Error Analysis Compare to linear spherical harmonic approximation n [Sloan, Kautz, Snyder, 2002] Measure approximation quality in two ways: n Error in lighting n Error in output image pixels Main point (for detailed natural lighting) n Non-linear wavelets converge exponentially faster than linear harmonics
44
Error in Lighting: St Peter’s Basilica Approximation Terms Relative L 2 Error (%) Sph. Harmonics Non-linear Wavelets
45
Error in Output Image: Plant Scene Approximation Terms Relative L 2 Error (%) Sph. Harmonics Non-linear Wavelets
46
Output Image Comparison Top: Linear Spherical Harmonic Approximation Bottom: Non-linear Wavelet Approximation 252002,00020,000
47
Summary A viable representation for all-frequency lighting n Sparse non-linear wavelet approximation n 100 – 1000 times information reduction Efficient relighting from detailed environment maps
48
Future Improvements Matrix compression n Transport matrices are very large (over 1 GB) n We only compress matrix rows n Can also compress columns (images) Non-linear coefficient selection n Area-weighted priority works best n Develop theory for how to approximate input vector for minimum error in output vector
49
Signal Processing Taxonomy of Lighting Effects n “All-frequency” vs “Exclusively low frequency”
50
Future All-Frequency Lighting Research n Realistic materials with changing view n Local lighting effects Gehry’s E.M.P. Museum
51
Acknowledgments
52
Relighting with Changing View Matrix formulation still works n Output vector contains exit radiance at scene vertices rather than image pixels Main difference: Must assume diffuse surfaces
53
Plant Reference
54
Teapot Reference
55
Lighting Error
56
Light-Transport Matrix Columns
57
Light-Transport Matrix
58
Matrix Row Wavelet Encoding
59
Extract Row
60
Matrix Row Wavelet Encoding Wavelet Transform
61
Matrix Row Wavelet Encoding Wavelet Transform
62
Matrix Row Wavelet Encoding Wavelet Transform
63
Matrix Row Wavelet Encoding Wavelet Transform
64
Matrix Row Wavelet Encoding Store back in matrix
65
From Frank Gehry Architecture, Ragheb ed. 2001 Focus on Shadows
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.