Specular manifold sampling for rendering high-frequency caustics and glints

Abstract

Scattering from specular surfaces produces complex optical effects that are frequently encountered in realistic scenes: intricate caustics due to focused reflection, multiple refraction, and high-frequency glints from specular microstructure. Yet, despite their importance and considerable research to this end, sampling of light paths that cause these effects remains a formidable challenge. In this article, we propose a surprisingly simple and general sampling strategy for specular light paths including the above examples, unifying the previously disjoint areas of caustic and glint rendering into a single framework. Given two path vertices, our algorithm stochastically finds a specular subpath connecting the endpoints. In contrast to prior work, our method supports high-frequency normal- or displacement-mapped geometry, samples specular-diffuse-specular ("SDS") paths, and is compatible with standard Monte Carlo methods including unidirectional path tracing. Both unbiased and biased variants of our approach can be constructed, the latter often significantly reducing variance, which may be appealing in applied settings (e.g. visual effects). We demonstrate our method on a range of challenging scenes and evaluate it against state-of-the-art methods for rendering caustics and glints.

A Multiscale Microfacet Model Based on Inverse Bin Mapping

“…Two regions of size 80×80 are selected from both scenes, and convergence plots are computed for each of them (middle and right column). We use the script provided by Zeltner et al [ZGJ20] to compute the plots.…”

“…As discussed in Section 2, we want to investigate the scenario of increasing the number of texels that fall within a typical image pixel. We do this by rendering the original scenes from Zeltner et al [ZGJ20], and by increasing the tiling of the input texture 8 times which results in increasing the texel-to-pixel ratio 8 2 times. The results from the comparisons are presented in Figure 7.…”

“…We compare our method against Yan et al [YHMR16], the current state of the art for rendering specular high-resolution normal maps. All comparison results in this subsection are based on the original code, scenes and scripts provided by Zeltner et al [ZGJ20], which includes the original implementation of Yan et al [YHMR16]. We also implemented our own method as a Mitsuba 2 BRDF [NVZJ19].…”

“…Practical disadvantages are a lack of support for area light sources, and typically at least minutes of pre-computation time that are needed for SH projections. Recently, specular manifold sampling (SMS) method was demonstrated to render glints with modest memory requirements and with similar convergence rates [ZGJ20]. The method also has a brief pre-computation: only a LEAN map [OB10] is built.…”

Optimised Path Space Regularisation

“…Techniques such as manifold next event estimation (MNEE) [HDF15] or, more recently, specular manifold sampling (SMS) [ZGJ20] can be used in classical path tracing to deterministically find intermediate vertices that obey Fermat's principle such as to form an admissible path for next event estimation. While MNEE only handles refraction through specular interfaces, SMS generalises the concept to allow the sampling of both reflected and refracted caustics in a unified framework.…”