Dissertation/Thesis Abstract

Facilitating the Acquisition of Realistic Material Appearance Models
by Lockerman, Yitzchak, Ph.D., Yale University, 2016, 126; 10584954
Abstract (Summary)

Over the last two decades, tools for rendering realistic three dimensional scenes have become available to anyone. Unfortunately, tools for acquiring realistic material appearance models to render have lagged behind. In this dissertation, we demonstrate a number of tools that provide low cost methods to capture these models. Particularly, we focus on two different aspects of appearance: the spatial variance encoded in texture and the subsurface scattering of light within an object.

Our first tool allows users to extract textures from arbitrary natural images. These images can be acquired from the web or captured by a camera. An interface then allows even a novice user to easily specify the minimum information needed to extract a desired texture. This tool is freely available as an online web application.

Next we provide a generalization of our first tool to allow for the extraction of all textures in an image at multiple levels of scale. In addition to creating more realistic textures, this allows the user to use this information in a number of novel ways to create new works of art.

Finally, we show that low cost consumer level equipment can be used to acquire the subsurface scattering properties of three dimensional objects. Additionally, we obtain geometric information from objects with strong subsurface scattering, a difficult challenge for most commercial shape acquisition systems.

We provide a discussion of future plans to continue our work to democratize material acquisition. This includes a design for a handheld version of our material acquisition system. We also discuss the possibility to apply our work to other fields such as medical imaging.

Indexing (document details)
Advisor: Rushmeier, Holly
School: Yale University
School Location: United States -- Connecticut
Source: DAI-B 78/07(E), Dissertation Abstracts International
Subjects: Computer science
Keywords: Computer Graphics, Diffusion Methods, Material Acquisition, Nonnegative Matrix Factorization, Texture Extraction, Visual Appearance Models
Publication Number: 10584954
ISBN: 978-1-369-63539-3
Copyright © 2020 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy