View-dependent precomputed light transport using non-linear Gaussian function approximations

Abstract

We propose a real-time method for rendering rigid objects with complex view-dependent effects under distant all-frequency lighting. Existing precomputed light transport approaches can render rich global illumination effects, but high-frequency view-dependent effects such as sharp highlights remain a challenge. We introduce a new representation of the light transport operator based on sums of Gaussians. The non-linear parameters of the representation allow for 1) arbitrary bandwidth because scale is encoded as a direct parameter; and 2) high-quality interpolation across view and mesh triangles because we interpolate the average direction of the incoming light, thereby preventing linear cross-fading artifacts. However, fitting the precomputed light transport data to this new representation requires solving a non-linear regression problem that is more involved than traditional linear and non-linear (truncation) approximation techniques. We present a new data fitting method based on optimization that includes energy terms aimed at enforcing good interpolation. We demonstrate that our method achieves high visual quality for a small storage cost and fast rendering time.