Title: A SignalProcessing Framework for Forward and Inverse Rendering
1A Signal-Processing Framework for Forward and
Inverse Rendering
Columbia University Feb 11, 2002
2Illumination Illusion
- People perceive materials more easily under
natural illumination than simplified
illumination.
Images courtesy Ron Dror and Ted Adelson
3Illumination Illusion
- People perceive materials more easily under
natural illumination than simplified
illumination.
Images courtesy Ron Dror and Ted Adelson
4Material Recognition
Photographs of 4 spheres in 3 different lighting
conditions courtesy Dror and Adelson
5Estimating BRDF and Lighting
Photographs
Geometric model
6Estimating BRDF and Lighting
Forward RenderingAlgorithm
Photographs
BRDF
Rendering
Lighting
Geometric model
7Estimating BRDF and Lighting
Forward RenderingAlgorithm
Photographs
BRDF
Novel lighting
Rendering
Geometric model
8Inverse Problems Difficulties
Ill-posed (ambiguous)
9Real-Time Rendering
- Interactive rendering with natural lighting,
physical BRDFs
10Motivation
- Understand nature of reflection and illumination
- Applications in computer graphics
- Real-time forward rendering
- Inverse rendering
-
11Contributions of Thesis
- Formalize reflection as convolution
- Signal-processing framework
- Practical forward and inverse algorithms
12Outline
- Motivation
- Reflection as Convolution
- Preliminaries, assumptions
- Reflection equation, Fourier analysis (2D)
- Spherical Harmonic Analysis (3D)
- Signal-Processing Framework
- Applications
- Summary and Implications
13Assumptions
14Assumptions
- Known geometry
- Convex curved surfaces no shadows,
interreflection
Complex geometry use surface normal
15Assumptions
- Known geometry
- Convex curved surfaces no shadows,
interreflection - Distant illumination
Illumination Grace Cathedral courtesy Paul
Debevec
Photograph of mirror sphere
16Assumptions
- Known geometry
- Convex curved surfaces no shadows,
interreflection - Distant illumination
- Homogeneous isotropic materials
Anisotropic
Isotropic
17Assumptions
- Known geometry
- Convex curved surfaces no shadows,
interreflection - Distant illumination
- Homogeneous isotropic materials
- Later, practical algorithms relax some
assumptions
18Reflection
19Reflection as Convolution (2D)
L
B
20Reflection as Convolution (2D)
21Reflection as Convolution (2D)
22Convolution
u
Signal f(x)
Output h(u)
Filter g(x)
23Convolution
u1
u
x
Signal f(x)
Output h(u)
Filter g(x)
24Convolution
u2
u
x
Signal f(x)
Output h(u)
Filter g(x)
25Convolution
u3
u
x
Signal f(x)
Output h(u)
Filter g(x)
26Convolution
u
x
Signal f(x)
Output h(u)
Filter g(x)
27Reflection as Convolution (2D)
Fourier analysis
R. Ramamoorthi and P. Hanrahan Analysis of
Planar Light Fields from Homogeneous Convex
Curved Surfaces under Distant Illumination SPIE
Photonics West 2001 Human Vision and Electronic
Imaging VI pp 195-208
28Related Work
- Qualitative observation of reflection as
convolution Miller Hoffman 84, Greene
86, Cabral et al. 87,99 - Reflection as frequency-space operator DZmura
91 - Lambertian reflection is convolution Basri
Jacobs 01 - Our Contributions
- Explicitly derive frequency-space convolution
formula - Formal quantitative analysis in general 3D case
29Spherical Harmonics
0
1
2 . . .
-1
-2
0
1
2
30Spherical Harmonic Analysis
2D
3D
31Outline
- Motivation
- Reflection as Convolution
- Signal-Processing Framework
- Insights, examples
- Well-posedness of inverse problems
- Applications
- Summary and Implications
32Insights Signal Processing
- Signal processing framework for reflection
- Light is the signal
- BRDF is the filter
- Reflection on a curved surface is convolution
33Insights Signal Processing
- Signal processing framework for reflection
- Light is the signal
- BRDF is the filter
- Reflection on a curved surface is convolution
Filter is Delta function Output Signal
34Insights Signal Processing
- Signal processing framework for reflection
- Light is the signal
- BRDF is the filter
- Reflection on a curved surface is convolution
Signal is Delta function Output Filter
35Phong, Microfacet Models
Mirror
Illumination estimation ill-posed for rough
surfaces
Analytic formulae in R. Ramamoorthi and P.
Hanrahan A Signal-Processing Framework for
Inverse Rendering SIGGRAPH 2001 pp 117-128
36Lambertian
Incident radiance (mirror sphere)
Irradiance (Lambertian)
37Inverse Lighting
Given B,? find L
- Well-posed unless denominator vanishes
- BRDF should contain high frequencies Sharp
highlights - Diffuse reflectors low pass filters Inverse
lighting ill-posed
38Inverse BRDF
Given B,L find ?
- Well-posed unless Llm vanishes
- Lighting should have sharp features (point
sources, edges) - BRDF estimation ill-conditioned for soft lighting
Area source Same BRDF
Directional Source
39Factoring the Light Field
- Light Field can be factored
- Up to global scale factor
- Assumes reciprocity of BRDF
- Can be ill-conditioned
- Analytic formula derived
Given B find L and ?
More knowns (4D) than unknowns (2D/3D)
40Outline
- Motivation
- Reflection as Convolution
- Signal-Processing Framework
- Applications
- Forward rendering (convolution)
- Inverse rendering (deconvolution)
- Summary and Implications
41Computing Irradiance
- Classically, hemispherical integral for each
pixel - Lambertian surface is like low pass filter
- Frequency-space analysis
Incident Radiance
Irradiance
429 Parameter Approximation
Order 0 1 term (constant)
Exact image
RMS error 25
439 Parameter Approximation
Order 1 4 terms (linear)
Exact image
RMS Error 8
449 Parameter Approximation
Order 2 9 terms (quadratic)
Exact image
RMS Error 1
For any illumination, average error Jacobs 01
45Comparison
Irradiance map Texture 256x256 Hemispherical Inte
gration 2Hrs
Irradiance map Texture 256x256 Spherical
Harmonic Coefficients 1sec
Incident illumination 300x300
46Video
R. Ramamoorthi and P. Hanrahan An Efficient
Representation for Irradiance Environment Maps
SIGGRAPH 2001 pp 497-500 R. Ramamoorthi and P.
Hanrahan Frequency Space Environment Map
Rendering submitted
47Video
48Inverse Rendering
Lighting
Unknown
Known
Known
BRDF
Unknown
Textures are a third axis
49Contributions
- Complex illumination
- Factorization of BRDF, lighting (find both)
- New representations and algorithms
- Formal study of inverse problems (well-posed?)
50Complications
- Incomplete sparse data (few photographs)
- Concavities Self Shadowing
- Spatially varying BRDFs
51Complications
- Challenge Incomplete sparse data (few
photographs) Difficult to compute
frequency spectra - Solution
- Use parametric BRDF model
- Dual angular and frequency space representation
52Algorithm Validation
Photograph
True values
53Algorithm Validation
Photograph
Renderings
Image RMS error 5
Known lighting
Unknown lighting
True values
54Inverse BRDF Spheres
Photographs
Renderings (Recovered BRDF)
55Complications
- Challenge Complex geometry with concavities
Self shadowing - Solution
- Use associativity of convolution
- Blur lighting, treat specular BRDF term as mirror
- Single ray for shadowing, easy in ray tracer
56Complex Geometry
- 3 photographs of a sculpture
- Complex unknown illumination
- Geometry known
- Estimate microfacet BRDF and distant lighting
57Comparison
58New View, Lighting
Photograph
Rendering
59Complications
- Challenge Spatially varying BRDFs
- Solution
- Use textures to modulate BRDF parameters
60Textured Objects
Rendering
Photograph
61Summary
- Reflection as convolution
- Frequency-space analysis gives many insights
- Practical forward and inverse algorithms
- Signal-Processing A useful paradigm for forward
and inverse rendering in graphics and vision
62Implications and Future Work
- Duality between forward and inverse problems
Ill-posed inverse problem ? fast forward
algorithm - Example Inverse lighting from Lambertian
surface ill-posed ? computing irradiance is fast
63Implications and Future Work
- Differential framework for reflection
- Analysis on object surface
- Complex illumination and BRDF
-
64Implications and Future Work
- Analyzing intrinsic structure, complexity of
light field Sampling theory based on
signal-processing - How many images in image-based rendering?
- How many principal components in PCA?
65Lighting Invariant Recognition
- Theory Space of images infinite-dimensional
for Lambertian Belhumeur and Kriegman 98
- Empirical 5D subspace enough for diffuse
objects
Hallinan 94, Epstein et al. 95, BK98,
Images from Yale face database
66Implications and Future Work
- Complex illumination in computer vision
- Generally assume simple lighting (point source)
without considering visibility (attached shadows) - Signal processing can be used to reduce effects
of complex illumination (with shadows) to
low-dimensional subspace - Many applications stereo, photometric stereo,
shape from shading, lighting invariant
recognition etc. -
67Acknowledgements
- Pat Hanrahan
- Marc Levoy
- Szymon Rusinkiewicz
- Steve Marschner
- Stanford graphics group
- Hodgson-Reed Stanford Graduate Fellowship
- NSF ITR grant 0085864 Interacting with the
Visual World
68Papers
- R. Ramamoorthi and P. Hanrahan A
Signal-Processing Framework for Inverse
Rendering SIGGRAPH 2001 pp 117-128 - R. Ramamoorthi and P. Hanrahan An Efficient
Representation for Irradiance Environment Maps
SIGGRAPH 2001 pp 497-500 - R. Ramamoorthi and P. Hanrahan Frequency Space
Environment Map Rendering submitted - R. Ramamoorthi and P. Hanrahan On the
Relationship between Radiance and Irradiance
Determining the Illumination from images of a
Convex Lambertian Object Journal of the Optical
Society of America A 18(10) 2001 pp 2448-2459 - R. Ramamoorthi and P. Hanrahan Analysis of
Planar Light Fields from Homogeneous Convex
Curved Surfaces under Distant Illumination SPIE
Photonics West 2001 Human Vision and Electronic
Imaging VI pp 195-208 - R. Ramamoorthi Analytic PCA Construction for
Theoretical Analysis of Lighting Variability,
Including Attached Shadows, in a Single Image of
a Convex Lambertian Object CVPR 2001 workshop on
Identifying Objects across Lighting Variations pp
48-55
ravir_at_graphics.stanford.edu http//graphics.stanf
ord.edu/ravir
69The End
70Photorealistic Rendering
71Measuring Materials, Light
Measure BRDF (reflectance) Point light source
72Interactive Forward Rendering
- Classically, rendering with natural
illumination is very expensive compared to using
simplified illumination
Directional Source
Natural Illumination
73Lighting Invariant Recognition
- Theory Infinite number of light directions
Space of images
infinite-dimensional - Empirical 5D subspace enough for diffuse objects
Images from Debevec et al. 00
74Lighting Invariant Recognition
- Theory Space of images infinite-dimensional
for Lambertian Belhumeur and Kriegman 98
- Empirical 5D subspace enough for diffuse
objects
Hallinan 94, Epstein et al. 95, BK98,
75Open Questions
- Relationship between spherical harmonics, PCA
- 9D approximation 5D empirical subspace
- Key insight Consider approximations over visible
normals (upper hemisphere), not entire sphere
Ramamoorthi CVPR IOAVL 01
76Light Field in 3D
- In flatland, 2D function
- In three dimensions, 4D function
-
Plenoptic Light Field
Surface Light Field
77 Dual Representation
- Diffuse BRDF Filter width small in frequency
domain - Specular Filter width small in spatial (angular)
domain - Practical Representation Dual angular,
frequency-space
78Related Work
- Precomputed (prefiltered) Irradiance maps
Miller and Hoffman 84, Greene 86, Cabral
et al 87 - Empirical observation Irradiance varies slowly
with surface normal. Use low resolution
irradiance maps - Contributions
- Analytic Irradiance formula
- Fast computation
- Compact 9 parameter representation
- Procedural rendering with programmable shading
hardware - Our approach can be extended to general BRDFs
79Comparison
Rendering (known L)
Photograph
Rendering (unknown L)