A BRDF analysis of cloth.

摘要

This thesis presents documentation and analysis of the Bidirectional Reflectance Distribution Function (BRDF) of cloth through actual measured data. Although significant cloth appearance modeling research exists, light reflectance has not been a focus. For many graphics applications, replicating the motion and/or texture of cloth coupled with a general reflectance model has been sufficient. For physical accuracy, however, correctly modeling how light interacts with a material is a necessity. Part of the difficulty in developing and verifying physically accurate models is scarcity of data. This thesis presents and examines reflectance data covering a breadth of cloth materials acquired with the UCSD Spherical Gantry. Over 30 different types of cloth were measured in total, made of some of the most common fibers such as cotton, linen, rayon, silk, wool, nylon, and polyester. Included in the analysis is an assessment of several well-known BRDF models that could be easily assumed to work for a material such as cloth. The performance of each model is evaluated with a non-linear constrained SQP algorithm that fits the models to the measured data. Through numerical errors and function plots, the conclusion is that while some models are sufficient for some cloth materials, most textiles exhibit complex reflectance properties that are not accounted for. Furthermore, BRDF data can vary greatly within the family of textiles depending on numerous variables such as fiber class, thread structure, fabrication method, and finishing.