Bounding the Albedo of the Ward Reflectance Model

is bounded by 1. Over the last five decades numerous BRDF models were introduced. Beckmann [BS63] and Torrance and Sparrow [TS67] presented physically based microfacet BRDF models that use the Gaussian distribution to define the microfacets’ surface normals. The model by Torrance and Sparrow was used in computer graphics by Cook and Torrance [CT81] and later improved by He et al. [HTSG91]. However, these models are neither suitable for Monte Carlo integration due to missing efficient importance sampling formulae, nor do they provide anisotropic reflection. The first empirical and probably most famous model that simulates specular reflections was introduced by Phong [Pho75] and later improved by Blinn [Bli77]. Other physically plausible BRDFs that model anisotropic reflection and are suitable for Monte Carlo integration were proposed by Schlick [Sch94], Lafortune et al. [LFTG97], Ashikmin and Shirley [AS00], and Kurt et al. [KSKK10]. As a simplification of the Cook-Torrance model, Ward [War92] presented an anisotropic BRDF that was later improved by Dür [Dür06]. The main benefits of this model are that it is computationally cheap to evaluate, it admits efficient importance sampling for Monte Carlo integration, and it