Revisiting the Foundations of Subsurface Scattering

Despite the significant advances in rendering, we are far from being able to automatically generate realistic and predictable images of organic materials such as plant and human tissues. Creating convincing pictures of these materials is usually accomplished by carefully adjusting rendering parameters. A key issue in this context is the simulation of subsurface scattering. Current algorithmic models usually rely on scattering approximations based on the use of phase functions, notably the Henyey-Greenstein phase function and its variations, which were not derived from biophysical principles of any organic material, and whose parameters have no biological meaning. In this report, we challenge the validity of this approach for organic materials. Initially, we present an original chronology of the use of these phase functions in tissue optics simulations, which highlights the pitfalls of this approach. We then demonstrate that a significant step toward predictive subsurface scattering simulations can be given by replacing it by a more efficient and accurate data oriented approach. Our investigation is supported by comparisons involving the original measured data that motivated the application of phase functions in tissue subsurface scattering simulations. We hope that the results of our investigation will help strengthen the biophysical basis required for the predictive rendering of organic materials.