Notes

Outline

CS 563 Advanced Topics in Computer Graphics Virtual Gonioreflectometry  by Guy Mann

Gonioreflectometer Used for determining the BRDF of a material Consists of a light source, an object being measured and a detector A defined amount of light is emitted at the object Light interacts with the surface Light is measured after its interaction with the object BRDF is determined from the amount and wavelengths of light detected after the interaction

Light Source Designed to give an even irradiance over a large area on the sample. The source is an quartz-halogen lamp with integral faceted ellipsoidal reflector. Light is focused on an opal glass diffuser to depolarize it, then collected by an aspheric condenser lens. An adjustable iris aperture is used as a spatial filter to even the illumination; this is placed at the focal plane of a Nikon 135mm camera lens, focused at infinity, which relays the light to the sample. Mounted on a 3-foot optical rail which can rotate across a range of 180 degrees to illuminate the sample from any angle. The source optics are fairly complex in order to meet the goals of good collimation, lack of polarization bias, even illumination across the sample, and high output.

Detector Light scattered from the sample passes through a dichroic polarizer and is reflected 90 degrees by a front-surface folding mirror, and is focused onto the detector slit by a 120mm achromatic doublet lens. A spectroradiometer serves as the detector; this contains a 1200 l/mm grating and acquires 1024 samples across the approximate range 385nm-710nm. The detector uses a diffraction grating to separate wavelengths. We measure the BRDF twice with polarizations 90 degrees apart, then use a weighted average to get the BRDF for a polarization-insensitive sensor.

Why Virtual? Simulates the interactions of gonioreflectometry; given geometry and the measurements of appearance (BRDF) Gathers reflectance information Source: Light distributed using Monte Carlo approximation Object: Geometry (micro-, milli-, object), BRDF Detector: Capture Sphere, Spherical Harmonics Render: this technique is used in Ray Tracers

Measuring  Scattered Rays Spherical Harmonics [1] Defines a function which represents the ray scattering, approximate BRDF, for a point on an object Capture Sphere [2] Defines a tree, representing facets of a geodesic sphere, which stores the ray scattering information

Spherical Harmonics A spherical harmonic function represents the scattering pattern of light from the geometry and the BRDF

Spherical Harmonics

Monte Carlo for Spherical Harmonics

Monte Carlo: Coefficients

Monte Carlo: Matrix

Capture Spheres Adaptively built Geodesic Sphere Each facet of the sphere is used as a bin to store the ratio of exitant to incident flux density Facets subdivide as needed Used to decide where to sample during rendering Cells with higher flux ratios are sampled more densely during rendering

Updating Capture Spheres Subdivision is based on the root mean squared deviation (RMSD) at each wavelength for the spectral flux density for each possible subfacet of a facet Basically: A facet is broken into its 4 possible subfacets and each one is tested for its spectral flux density.  If the RMSD is above a small tolerance then the facet is subdivided.

Using Scale The scattering representation for a smaller scale can be used as the BRDF for a larger scale

Microscale to Object Scale Microgeometry of Velvet The microgeometry sampled by a virtual gonioreflectometer. The output, a bidirectional reflectance distribution function or BRDF, can be used to generate images of velvet objects."

Isotropic vs. Anisotropic Isotropic Surfaces: Light interacts independently of the direction of incidence Anisotropic Surfaces: Light interacts with the surface differently depending on the direction of incidence Anisotropic surfaces can be created by using milliscale geometry

References Westin S. H., J.R. Arvo, K.E. Torrance, “Predicting Reflectance Functions from Complex Surfaces”, Proc. ACM SIGGRAPH 1992 Gondek J.S., Meyer, G.W., and Newman J.G, “Wavelength Dependent Reflectance Functions”, Proc. ACM SIGGRAPH 1994 Cabral B. et al, “Bidirectional Reflection Functions from Surface Bump Maps”, Proc. ACM SIGGRAPH 1987, pp 273-281 Meyer G W et al, “A Computer Graphics System for Rendering Gonio-Apparent Colors”, Proc. AIC Congress 2001 Cornell University Program of Computer Graphics, Cornell Gonioreflectometer, [Online], Available: http://www.graphics.cornell.edu/research/measure/gonio.html [February 2003] Imager Computer Graphics Laboratory, The Main Image Gallery, [Online] Available: http://www.cs.ubc.ca/nest/imager/contributions/bobl/imagergallery/main/top.html, [February 2003]