group photo

Make us dance!

6th International Radiance and HDR Scientific Workshop

Monday, October 1 - Tuesday, October 2, 2007

Meeting Schedule

October 1
8:30–9:30 Meet & Greet / Continental Breakfast
9:30–9:45 Greg Ward & Katja Doerschner, Introduction
Roland Fleming, Max Planck Institute for Biological Cybernetics, Tuebingen, German

"Visual Perception of Surface Material"
Different materials such as silk, bronze and marmalade have distinctive visual appearances. Human observers are remarkably adept at recognizing materials across a wide range of viewing conditions and we are only just beginning to work out how. What gives a material its characteristic 'look'? What cues does the visual system use to identify materials? How can we leverage the assumptions made by the visual system to improve computer graphics? I will review some of our research on the perception of material properties such as gloss, translucency and refractive index that attempts to shed some light on these questions. Ill talk about how illumination and 3D geometry interact with material perception, and discuss the role of various image statistics (e.g. intensity histogram, amplitude spectrum) in the visual estimation of material attributes. Ill then show one application in which we exploit the heuristics made by the visual system to enable illusory modifications of material appearance in photographs.
10:30–10:50 Coffee Break
Yu Sheng, Rensselaer Polytechnic Institute, Troy, NY

"Comparing an interactive hybrid global illumination method with Radiance"
Complex fenestration systems (CFS) can be used to redirect intense illumination from the sun to more evenly illuminate an architectural space and thus reduce the lighting needs of buildings. However, standard distribution of Radiance does not support rendering of CFS or arbitrary BRDF/BTDF material data. In this talk, we present an interactive rendering system for architectural design which simulates a CFS based on 4D Bidirectional Transmission Distribution Function (BTDF) data. We use a hybrid method of shadow volumes for direct illumination and radiosity for indirect illumination to achieve interactive rendering rates.

Our system is appropriate for use in schematic design: an early stage of the architectural design process where scale, appearance, and adjacencies of an evolving design are explored. We demonstrate our system on several models inspired by field observations and the designs of architecture students we have consulted during the development of this project.
Bei Xiao, University of Pennsylvania, Philadelphia, PA

"Effect of test patch location on color appearance, in the context of 3D objects"
Bei Xiao (1) and David Brainard (2)
(1) Department of Neuroscience, School of Medicine, University of Pennsylvania, Philadelphia, PA, 19104, USA. (2) Department of Psychology, University of Pennsylvania, Philadelphia, PA, 19104, USA.

The light reflected from different locations on a single object can vary enormously. This variation is enhanced when the material properties of the object are changed from matte to glossy. Yet humans have no trouble perceiving an object as having a unified color. We conducted a psychophysical experiment to study how people perceive the color of different parts of an object. Observers viewed graphics simulations of a three-dimensional scene containing two objects, test and match (generated with Radiance in conjunction with custom software that provided full spectral rendering). The test object was a soccer ball that had one colored hexagonal face (test patch). Observers were asked to adjust the color appearance of a match sphere to the test patch. The match sphere was always matte, while we varied the surface gloss of the entire soccer ball (including the test patch). The test patch could be located at either an upper or lower location on the soccer ball. The data show that there is an effect of test patch location on observers' color matching, but this effect is small compared to the physical change in the average light reflected from the test patch across the two locations. In addition, we found that observers exhibit stability of color perception of the test patch in the face of variation of surface gloss.
12:00–1:30 Lunch
1:30–2:30 Gary Meyer, Victoria Interrante, University of Minnesota, Department of Computer Science

"Digital Design Consortium and VR lab tour/demos"
Greg Ward, Anyhere Software, Albany, CA

Tutorial: "Image-based Lighting"
Image-based lighting (IBL) is a set of techniques originated by Paul Debevec for incorporating synthetic objects into real-world scenery IBL is widely used by the movie industry for special effects production, and offers some interesting possibilities for daylight simulation as well. In this tutorial, Greg will demonstrate how easy it is to capture a high dynamic-range environment map (a.k.a. a "light probe") using a standard digital camera and use it to render synthetic objects into a background plate with the help of Photosphere and the "mksource" program. We will discuss how other researchers, such as Santiago Torres, have employed IBL and Radiance for daylighting simulation.
3:15–3:30 Coffee Break
Zack Rogers, Architectural Energy Corporation, Boulder, CO

Tutorial: "Sensor Placement + Optimization Tool (SPOT) Update"
This talk will present recent updates to the Sensor Placement + Optimization Tool software that acts as a front-end interface to Radiance. SPOT has grown to provide more detailed electric lighting and annual daylighting analysis as well as providing photosensor control analysis. The focus of the presentation will be on the new features of the software and a case study will be presented of a project in which we used SPOT to help design a daylight responsive control system. The talk will be 25 minutes followed by a question and answer period.
6:30–7:30 Reception
7:30–10:30 Banquet


October 2:
8:30–9:00 Continental Breakfast
9:00–10:00 Zack Rogers, Architectural Energy Corporation, Boulder, CO

"Experiences with Radiance in Daylighting Design, Part III"
Zack Rogers, Galen Burrell and Jennifer Scheib, Architectural Energy Corporation

This talk will present some of the Radiance modeling Architectural Energy Corporation has done in support of our Daylighting Design Consulting work, focusing on the various ways we have used Radiance to guide the design process. Radiance has proved to be extremely effective in analyzing and visualizing daylighting designs, allowing numerous daylighting design alternatives to be explored beforehand, informing and guiding the daylighting design decisions of a project. The types of daylighting design projects that will be highlighted vary widely and include schools, laboratories, offices, museums, atriums, and others. The focus of the presentation will be on new work that has occurred since the Montreal radiance conference in 2005.

The talk will be broken into three-20 minute sections, one section given by each of the presenters followed by a brief question and answer period.
10:00–10:25 Coffee break
Greg Ward, Anyhere Software, Albany, CA

"Utilizing BTDF Window Data"
This talk presents some new ideas and software for exploiting measured and simulated bidirectional transmittance distribution function (BTDF) data in Radiance. Programs and standards for providing BTDF data are emerging (e.g., LBNL Window 6), and we wish to take advantage of it in our simulations. Two approaches will be presented, one geared towards traditional daylight simulation and rendering and the other towards annual calculations. The first approach resembles mkillum but without the usual restrictions, since BTDFs can represent virtually any complex fenestration type. The second approach requires two passes of rtcontrib, one to account for the building exterior, and the other for the interior. This fully separates illumination from geometry and permits time-based fenestration controls to be simulated efficiently. As this constitutes work in progress, suggestions will be most appreciated.
Daniel Lichtman, University of Pennsylvania, Philadelphia, PA

"RenderToolbox: A MATLAB Toolkit for Hyperspectral Rendering with Radiance and PBRT"
Daniel P. Lichtman, Bei Xiao, David H. Brainard

We describe a set of Matlab software tools, the RenderToolbox, that aid in the modeling and rendering of images for use in psychophysical experiments. The toolbox, which we are making freely available, has several important features. First, it allows the user to model scenes in the popular Maya software package and export these into the remainder of the rendering pipeline. Second, although the toolbox inherits the scene geometry from the Maya modeler, it allows the user to associate full spectral reflectance functions and parametric BRDFs with each object, and full spectral power distributions with each illuminant. The toolbox then parses the scenes and associated reflectance/illuminant parameters and passes these to either of two renderers. These are Radiance and Physically Based Rendering Tools (PBRT). It invokes the renderers on a wavelength-by-wavelength basis to produce a hyperspectral image of the scene. The fact that the toolbox transparently supports two renderers allows easy comparison of their performance. In addition, the toolbox is configured to make it easy to re-render the same scene geometry with different choices of reflectance/illuminant parameters. Fourth, the package provides support for converting the hyperspectral image to standard color representations. Finally, the package supports parallel rendering of the separate wavelength images if a computer cluster is available. Several example scenes are included with the toolbox to demonstrate its use and to compare the two renderers' performance. The simplest example, a uniformly reflective surface and a single point-light source, yields two nearly identical images. Each of these matches an analytical prediction based on the light's spectral power distribution and surface's reflectance function. A second example scene includes a single sphere with uniform reflectance under a distant point-light source. Again, images rendered by Radiance and PBRT are well-matched to each other and to the directly calculated image based on the Ward model of surface reflectance. The close agreement of the two renderers with each other and with direct calculations for simple scenes provides added confidence that each renderer is doing a good job simulating physical light flow.
Susan Ubbelohde, Loisos + Ubbelohde Associates, Oakland, CA

"Daylighting in Practice: Radiance and the Design Process"
Loisos + Ubbelohde is a consulting and design firm specializing in sustainable architecture. Our consulting work focuses on advanced daylighting design and performance, energy modeling and alternative energy sources. In practice, the use of tools such as Radiance and physical models is never as direct and linear as theory might assume. Rather, our use of design and evaluation tools is continually adapted to the design direction of the project, the questions posed by the architects, the tradeoffs and negotiations necessary with mechanical, structural, electrical engineers, lighting designers and interior designers on the design team, and the goals and concerns of the building owners and the requirements of outside agencies such as the US Green Building Council. We will present a range of recent projects, both consulting projects and design projects, discussing our use of modeling tools in the projects.
12:00–1:30 Lunch
Greg Ward, Anyhere Software, Albany, CA

Tutorial: "Improved Color Rendering with RGB"
Accurate color rendering requires the consideration of many samples over the visible spectrum, and advanced rendering tools developed by the research community offer multispectral sampling towards this goal. However, for practical reasons including efficiency, white balance, and data demands, Radiance still employs a simple RGB model in its lighting calculations. Applied naively, this can result in colors that are qualitatively different from the correct ones. In this tutorial, we demonstrate two independent and complementary techniques for improving RGB rendering accuracy in Radiance without impacting calculation time: spectral prefiltering and color space selection. Spectral prefiltering is an obvious but overlooked method of preparing input colors for a conventional RGB rendering calculation, which achieves exact results for the direct component, and very accurate results for the interreflected component when compared with full-spectral rendering. In particular, we demonstrate the merits of a particular color space transform that has emerged from the research community as the best performer in computing white point adaptation under changing illuminants: the Sharp RGB space.
2:30–2:45 Coffee
Katja Doerschner, Department of Psychology, University of Minnesota, Minneapolis, MN

"Perceived Shininess and Rigidity — Measurements of Shape-Dependent Specular Flow of Rotating Objects"
Katja Doerschner (1), Paul Schrater (1,2), Dan Kersten (1) (1) Department of Psychology, University of Minnesota , (2) Department of Computer Science, University of Minnesota

Specular flow provides useful constraints when estimating an object's shape (Roth & Black 2006, Fleming et al. 2003). We tested experimentally whether human observers can exploit these constraints when judging the rigidity and shininess of rotating specular superellipsoids of varying corner-roundedness. Our results indicate that observers perceived more-rounded shapes (ellipsoids) less rigid and less shiny than cuboidal shapes. We currently explore whether observers' perceptions correlate with the characteristics of the image velocity distribution that these rotating objects generate.