Two-Stage Optimization for Fitting Parameters of Measured Isotropic Spectral BRDFs

Parametric Bidirectional Reflectance Distribution Function (BRDF), which represents the surface reflectance with parametric models, is compact, efficient to render, and easy to edit. The parameters for the parametric BRDFs are often obtained by fitting them with measured surface reflectances of real-world materials at RGB channels. In recent years, the surface reflectances at densely sampled wavelengths, referred to as measured spectral BRDFs, have been publicized. Fitting parametric BRDFs with measured spectral BRDFs is difficult since the number of parameters is prohibitive compared with that for measured RGB BRDFs. This paper proposes a two-stage optimization method to fit the parameters of parametric BRDFs with measured spectral BRDFs. In the first stage, our method fits the wavelength-independent parameter with the average surface reflectance of all discretized wavelengths. Then, the wavelength-dependent parameters are fitted for each discretized wavelength, reducing the number of simultaneously optimized parameters. Experimental results show that our fitting parameters can provide plausible rendering results.