Efficient Rendering of Heterogeneous Polydisperse Granular Media

摘要

We address the challenge of efficiently rendering massive assembliesrnof grains within a forward path-tracing framework. Previousrnapproaches exist for accelerating high-order scattering for a limited,rnand static, set of granular materials, often requiring scene-dependentrnprecomputation. We significantly expand the admissible regime ofrngranular materials by considering heterogeneous and dynamic granularrnmixtures with spatially varying grain concentrations, pack rates,rnand sizes. Our method supports both procedurally generated grainrnassemblies and dynamic assemblies authored in off-the-shelf particlernsimulation tools. The key to our speedup lies in two complementaryrnaggregate scattering approximations which we introduced to jointlyrnaccelerate construction of short and long light paths. For low-orderrnscattering, we accelerate path construction using novel grain scatteringrndistribution functions (GSDF) which aggregate intra-grain lightrntransport while retaining important grain-level structure. For highorderrnscattering, we extend prior work on shell transport functionsrn(STF) to support dynamic, heterogeneous mixtures of grains withrnvarying sizes. We do this without a scene-dependent precomputationrnand show how this can also be used to accelerate light transport in arbitraryrncontinuous heterogeneous media. Our multi-scale renderingrnautomatically minimizes the usage of explicit path tracing to onlyrnthe first grain along a light path, or can avoid it completely, whenrnappropriate, by switching to our aggregate transport approximations.rnWe demonstrate our technique on animated scenes containing heterogeneousrnmixtures of various types of grains that could not previouslyrnbe rendered efficiently. We also compare to previous work on a simplerrnclass of granular assemblies, reporting significant computationrnsavings, often yielding higher accuracy results.