Comparison of measured and computed BSDF of a daylight redirecting component

Abstract

The Bidirectional Scatter Distribution Function (BSDF) of a selected Daylight Redirecting Component (DRC) is computed by a virtual goniophotometer using the enhanced photon map extension in Radiance, and compared to measured BSDF data. The DRC comprises a stack of tilted aluminum louvers with configurable inclination angle. The profile of the louvers is designed to control transmission depending on sun altitude, and to redirect light up towards the ceiling. The measured BSDF of the DRC is obtained from a scanning goniophotometer. For a sparse set of three source directions, the distribution is recorded at ≃ 250,000 receiver directions. The asymmetric angular resolution allows detailed observation of characteristic features in the distribution, which are assumed to persist over a range of source directions. For each pair of source and receiver directions in the measurement, the computed BSDF is generated from a model of the DRC, replicating the measurement with a virtual goniophotometer. The simulation relies only on the enhanced photon map extension for Radiance. The BSDF from measurement and simulation are compared qualitatively and quantitatively to discuss the degree of accordance. The presence of characteristic features and their topology is evaluated by comparing polar surface plots of the distributions and profiles of the scatter plane. The direct-hemispherical transmission is compared for each measurement and simulation. The RMSE of each computed distribution against the corresponding measurements is calculated to quantify the directionally resolved deviation. A high degree of qualitative accordance between the computed and the measured BSDF is achieved. Prominent features in the BSDF are represented by the model. A deviation of −6% to +15% is observed in a quantitative comparison of direct-hemispherical transmission by integration of computed and measured BSDF. The RMSE indicates higher deviations for lower source altitudes, where a direct transmission peak in the distribution is underestimated by the model. The method is proposed as a means to validate the capability of the enhanced photon map to predict transmission through DRC.