針對BRDF的全域光源重要性採樣方法之研究

Abstract

我們藉由球面輻射基底函數(SRBF)的特性，針對雙向反射分佈函數(BRDF)提出一種新的重要性採樣方法。我們經由實驗證實，透過非線性最佳化的演算法，將雙向反射分佈函數轉換至球面輻射基底函數的表示形式之誤差是在一個可容忍的誤差範圍內。此外，根據球面輻射基底函數的表現形式，我們也提出一種有效率的重要性採樣方法，其產生出來的採樣分佈可以契合球面輻射基底函數所描述的資料分佈，此方法可以在全域光源照射的環境下，有效率地繪製包含多種複雜材質的場景影像。

We present a new technique for sampling the importance of realistic bidirectional reflectance distribution function (BRDF) models by using Spherical Radial Basis Functions (SRBFs). We demonstrate that the measured BRDF data can be represented in SRBFs in a tolerated error bound by using our non-uniform and non-negative SRBF fitting algorithm. Then, we present an efficient sampling algorithm on the unit sphere, and the generated point distribution can match the SRBF representation. We are capable of creating images with multiple measured BRDFs under global illumination.