Addy Ngan

Experimental analysis of BRDF models

The Bidirectional Reflectance Distribution Function (BRDF) describes the appearance of a material by its interaction with light at a surface point. A variety of analytical models have been proposed to represent BRDFs. However, analysis of these models has been scarce due to the lack of high-resolution measured data. In this work we evaluate several well-known analytical models in terms of their ability to fit measured BRDFs. We use an existing high-resolution data set of a hundred isotropic materials and compute the best approximation for each analytical model. Furthermore, we have built a new setup for efficient acquisition of anisotropic BRDFs, which allows us to acquire anisotropic materials at high resolution. We have measured four samples of anisotropic materials (brushed aluminum, velvet, and two satins). Based on the numerical errors, function plots, and rendered images we provide insights into the performance of the various models. We conclude that for most isotropic materials physically-based analytic reflectance models can represent their appearance quite well. We illustrate the important difference between the two common ways of defining the specular lobe: around the mirror direction and with respect to the half-vector. Our evaluation shows that the latter gives a more accurate shape for the reflection lobe. Our analysis of anisotropic materials indicates current parametric reflectance models cannot represent their appearances faithfully in many cases. We show that using a sampled microfacet distribution computed from measurements improves the fit and qualitatively reproduces the measurements.

Image-based 3D photography using opacity hulls

We have built a system for acquiring and displaying high quality graphical models of objects that are impossible to scan with traditional scanners. Our system can acquire highly specular and fuzzy materials, such as fur and feathers. The hardware set-up consists of a turntable, two plasma displays, an array of cameras, and a rotating array of directional lights. We use multi-background matting techniques to acquire alpha mattes of the object from multiple viewpoints. The alpha mattes are used to construct an opacity hull. The opacity hull is a new shape representation, defined as the visual hull of the object with view-dependent opacity. It enables visualization of complex object silhouettes and seamless blending of objects into new environments. Our system also supports relighting of objects with arbitrary appearance using surface reflectance fields, a purely image-based appearance representation. Our system is the first to acquire and render surface reflectance fields under varying illumination from arbitrary viewpoints. We have built three generations of digitizers with increasing sophistication. In this paper, we present our results from digitizing hundreds of models.

Analysis of human faces using a measurement-based skin reflectance model

We have measured 3D face geometry, skin reflectance, and subsurface scattering using custom-built devices for 149 subjects of varying age, gender, and race. We developed a novel skin reflectance model whose parameters can be estimated from measurements. The model decomposes the large amount of measured skin data into a spatially-varying analytic BRDF, a diffuse albedo map, and diffuse subsurface scattering. Our model is intuitive, physically plausible, and -- since we do not use the original measured data -- easy to edit as well. High-quality renderings come close to reproducing real photographs. The analysis of the model parameters for our sample population reveals variations according to subject age, gender, skin type, and external factors (e.g., sweat, cold, or makeup). Using our statistics, a user can edit the overall appearance of a face (e.g., changing skin type and age) or change small-scale features using texture synthesis (e.g., adding moles and freckles). We are making the collected statistics publicly available to the research community for applications in face synthesis and analysis.

Modeling acoustics in virtual environments using the uniform theory of diffraction

Realistic modeling of reverberant sound in 3D virtual worlds provides users with important cues for localizing sound sources and understanding spatial properties of the environment. Unfortunately, current geometric acoustic modeling systems do not accurately simulate reverberant sound. Instead, they model only direct transmission and specular reflection, while diffraction is either ignored or modeled through statistical approximation. However, diffraction is important for correct interpretation of acoustic environments, especially when the direct path between sound source and receiver is occluded. The Uniform Theory of Diffraction (UTD) extends geometrical acoustics with diffraction phenomena: illuminated edges become secondary sources of diffracted rays that in turn may propagate through the environment. In this paper, we propose an efficient way for computing the acoustical effect of diffraction paths using the UTD for deriving secondary diffracted rays and associated diffraction coefficients. Our main contributions are: 1) a beam tracing method for enumerating sequences of diffracting edges efficiently and without aliasing in densely occluded polyhedral environments; 2) a practical approximation to the simulated sound field in which diffraction is considered only in shadow regions; and 3) a real-time auralization system demonstrating that diffraction dramatically improves the quality of spatialized sound in virtual environments.

Acquisition and rendering of transparent and refractive objects

This paper introduces a new image-based approach to capturing and modeling highly specular, transparent, or translucent objects. We have built a system for automatically acquiring high quality graphical models of objects that are extremely difficult to scan with traditional 3D scanners. The system consists of turntables, a set of cameras and lights, and monitors to project colored backdrops. We use multi-background matting techniques to acquire alpha and environment mattes of the object from multiple viewpoints. Using the alpha mattes we reconstruct an approximate 3D shape of the object. We use the environment mattes to compute a high-resolution surface reflectance field. We also acquire a low-resolution surface reflectance field using the overhead array of lights. Both surface reflectance fields are used to relight the objects and to place them into arbitrary environments. Our system is the first to acquire and render transparent and translucent 3D objects, such as a glass of beer, from arbitrary viewpoints under novel illumination.

Image-driven navigation of analytical BRDF models

Specifying parameters of analytic BRDF models is a difficult task as these parameters are often not intuitive for artists and their effect on appearance can be non-uniform. Ideally, a given step in the parameter space should produce a predictable and perceptually-uniform change in the rendered image. Systems that employ psychophysics have produced important advances in this direction; however, the requirement of user studies limits scalability of these approaches. In this work, we propose a new and intuitive method for designing material appearance. First, we define a computational metric between BRDFs that is based on rendered images of a scene under natural illumination. We show that our metric produces results that agree with previous perceptual studies. Next, we propose a user interface that allows for navigation in the remapped parameter space of a given BRDF model. For the current settings of the BRDF parameters, we display a choice of variations corresponding to uniform steps according to our metric, in the various parameter directions. In addition to the parametric navigation for a single model, we also support neighborhood navigation in the space of all models. By clustering a large number of neighbors and removing neighbors that are close to the current model, the user can easily visualize the alternate effects that can only be expressed with other models. We show that our interface is simple and intuitive. Furthermore, visual navigation in the BRDF space both in the local model and the union space is an effective way for reflectance design.

Statistical acquisition of texture appearance

We propose a simple method to acquire and reconstruct material appearance with sparsely sampled data. Our technique renders elaborate view- and light-dependent effects and faithfully reproduces materials such as fabrics and knitwears. Our approach uses sparse measurements to reconstruct a full six-dimensional Bidirectional Texture Function (BTF). Our reconstruction only require input images from the top view to be registered, which is easy to achieve with a fixed camera setup. Bidirectional properties are acquired from a sparse set of viewing directions through image statistics and therefore precise registrations for these views are unnecessary. Our technique is based on multi-scale histograms of image pyramids. The full BTF is generated by matching the corresponding pyramid histograms to interpolated top-view images. We show that the use of multi-scale image statistics achieves a visually plausible appearance. However, our technique does not fully capture sharp specularities or the geometric aspects of parallax. Nonetheless, a large class of materials can be reproduced well with our technique, and our statistical characterization enables acquisition of such materials efficiently using a simple setup.

Experimental validation of analytical BRDF models

Multiple Lobes Cook and Torrance suggest that two or more specular lobes are necessary to model surfaces where there are multiple scales of roughness. Lafortune’s model serve as basis functions for representing BRDFs and thus multiple lobes are also expected in general. To study the effect of an extra lobe, we choose 26 materials from the dataset, typically materials that have multiple layers of nishing, e.g. metallic paints. The tting errors with the Cook-Torrance and the Lafortune models with one and two lobes are shown in Figure 1. In 14 cases for Cook-Torrance and 11 cases for Lafortune, an extra lobe reduces the error by more than 20%. Again, we note that the Lafortune model perform worse than the Cook-Torrance model.

Processing and editing of faces using a measurement-based skin reflectance model

In extension to our paper Analysis of Human Faces using a Measurement-Based Skin Reflectance Model we describe the almost fully automated face processing pipeline including the custom-built acquisition system, calibration procedures, and rendering. The underlying model is intuitive, physically plausible, and allows easy face editing. Generated image quality approaches that of real photographs.