Eugene d’Eon | Steve Marschner | Johannes Hanika
SIGGRAPH Talks 2014


Analytic fiber scattering models are popular for rendering hair and fur. All of these models are the sum of distinct lobes, R, TT, TRT, and so on, that correspond to specific pathways that light takes as it interacts with the fiber. Previous models have proposed that these lobes are well represented by separable products of longitudinal and azimuthal functions. We revisit the derivation of farfield fiber scattering and find that non-separable behaviours arise. Just as the Phong reflectance model for surfaces lacks important lobe-tightening behaviour at grazing angles, we find analagous missing behaviours in previous fiber scattering models. We derive new practical analytic lobes that exhibit these non-separable behaviours and find these to closely match Monte Carlo ground truth. In the talk we will briefly highlight how the new lobes are derived and explore the application of these new lobes for rendering human hair.