Non-parametric modeling and editing for captured SVBRDF

Abstract

Reflectance properties of real-world objects can be represented by spatially varying bidirectional reflectance distribution function (SVBRDF). Non-parametric BRDF becomes the main aspect nowadays because of its verisimilitude and generality. We present a new method for capturing, modeling and editing non-parametric SVBRDF. Our method seeks to achieve high reconstruction accuracy, compactness and editability of representation, and meanwhile to speed up the SVBRDF modeling processes. For a planar surface, we (1) design a capturing device to acquire reflectance samples of the surface; (2) propose a Laplacian-based angular interpolation scheme for a 2D slice of BRDF at a given surface location, and a kernel Nyström method for SVBRDF data matrix reconstruction; (3) propose a practical algorithm to extract linearly independent basis BRDFs, and to calculate blending weights by projecting reconstructed reflectance onto these basis BRDFs; (4) decompose these basis BRDFs into groups of 1D curves for editing intuitively. Our experiment results demonstrate that our approach can model real-world reflectance with both high accuracy and high visual fidelity for real-time virtual environment rendering.