Subdivision next-event estimation for path-traced subsurface scattering

1Disney Research 2Walt Disney Animation Studios 3University of Stuttgart 4Dartmouth College

In Proceedings of EGSR (Experimental Ideas & Implementations), 2016

Equal-time comparison of standard unidirectional path tracing (left halves) against SNEE (right halves) using the same setup as in Figure 3. Here we show all transport, including the transport which is not handled by our technique, such as single scattering. SNEE is designed to handle well the difficult cases and provides most benefits when the BSDF is narrow and the illumination high-frequency.


We present subdivision next-event estimation (SNEE) for unbiased Monte Carlo simulation of subsurface scattering. Our technique is designed to sample high frequency illumination through geometrically complex interfaces with highly directional scattering lobes enclosing a scattering medium. This case is difficult to sample and a common source of image noise. We explore the idea of exploiting the degree of freedom given by path vertices within the interior medium to find two-bounce connections to the light that satisfy the law of refraction. SNEE first finds a surface point by tracing a probe ray and then performs a subdivision step to place an intermediate vertex within the medium according to the incoming light at the chosen surface point. Our subdivision construction ensures that the path will connect to the light while obeying Fermat's principle of least time. We discuss the details of our technique and demonstrate the benefits of integrating SNEE into a forward path tracer.



David Koerner, Jan Novák, Peter Kutz, Ralf Habel, Wojciech Jarosz. Subdivision next-event estimation for path-traced subsurface scattering. Proceedings of EGSR (Experimental Ideas & Implementations), June 2016.
    author    = {Koerner, David and Nov\'ak, Jan and Kutz, Peter and Habel, Ralf and Jarosz, Wojciech},
    title     = {Subdivision Next-Event Estimation for Path-Traced Subsurface Scattering},
    booktitle = {Proceedings of EGSR (Experimental Ideas \\& Implementations)},
    year      = {2016},
    month     = jun,
    publisher = {The Eurographics Association},
    doi       = {10/gf6rzj}
© The Author(s). This is the author's version of the work. It is posted here by permission of The Eurographics Association for your personal use. Not for redistribution. The definitive version is available at