Multi-scale modeling and rendering of granular materials

Abstract

85%10% 5% f = 0.63 Ours EPT 628 / 1736 hrs (4.8e-6) 50 / 789 hrs (4.9e-6) EPT (1.1e-4) Ours (4.8e-5) EPT (5.5e-5) 28/77 hrs 5/78 hrs 56/155 hrs Equal time Equal variance Figure 1: We propose a multi-scale procedural approach for modeling granular materials. The user specifies the bounding shape for the aggregate material (top left), selects a pre-packed tile of grain bounding spheres (top middle), within which we instantiate randomly rotated copies of the selected exemplar grains (bottom left) according to the specified mixing ratios. The SANDCASTLE contains about 2 billion grains, each composed of approximately 200 k triangles. We report the high-order / total render times in hours and the variance in parentheses. Our approach (top half) renders the high-order scattering over 12× (50 vs. 628 hrs) faster than explicitly path tracing (EPT) the individual grains (bottom half) while providing visually indistinguishable results. The insets on the right provide equal time and equal variance comparisons. AbstractWe address the problem of modeling and rendering granular materials-such as large structures made of sand, snow, or sugarwhere an aggregate object is composed of many randomly oriented, but discernible grains. These materials pose a particular challenge as the complex scattering properties of individual grains, and their packing arrangement, can have a dramatic effect on the large-scale appearance of the aggregate object. We propose a multi-scale modeling and rendering framework that adapts to the structure of scattered light at different scales. We rely on path tracing the individual grains only at the finest scale, and-by decoupling individual grains from their arrangement-we develop a modular approach for simulating longer-scale light transport. We model light interactions within and across grains as separate processes and leverage this decomposition to derive parameters for classical radiative transport, including standard volumetric path tracing and a diffusion method that can quickly summarize the large scale transport due to many grain interactions. We require only a one-time precomputation per exemplar grain, which we can then reuse for arbitrary aggregate shapes and a continuum of different packing rates and scales of grains. We demonstrate * The work was done while the author was employed at Disney Research.our method on scenes containing mixtures of tens of millions of individual, complex, specular grains that would be otherwise infeasible to render with standard techniques.

A Stochastic Film Grain Model for Resolution‐Independent Rendering

“…A promising direction is to build upon the recent work on the rendering of granular materials by Meng et al . [MPH*15].…”

A radiative transfer framework for spatially-correlated materials

“…Volumetric representation of appearance Volumetric representations of explicit geometry have been successfully used to approximate complex appearances. Meng et al [2015] used a classical radiative approximation of light transport in particulate media for efficient rendering. Fiber-level cloth appearance models based either on micro-CT geometry [Zhao et al 2011[Zhao et al , 2012 or procedural modeling [Schröder et al 2011], have used volumetric anisotropic representations for rendering high-detailed garments [Aliaga et al 2017], similar in quality to explicit fiber representations [Khungurn et al 2015].…”

“…However, the distribution of scatterers in many media exhibit different forms and degrees of spatial correlation (e.g. clouds [Davis and Marshak 2004;Lovejoy et al 1995], textiles [Coquard and Baillis 2006], porous materials [Bellet et al 2009;Taine et al 2010], or granular aggregates [Meng et al 2015]). This affects light transport, as Figure 4 illustrates; as a consequence, attenuation is no longer exponential, and light extinction becomes non-Poissonian.…”

“…In order to validate the existence of non-exponential transmittance, in addition to findings from other fields such as neutron transport or atmospheric sciences, we performed a simple experiment where we capture the transmittance of different correlated (non-exponential) and uncorrelated (exponential) media. For capture, we use a setup inspired in Meng et al [Meng et al 2015, Figure 3], where we filled a glass-made vase with the material. The vase was placed on top of a mobile flash for lighting, and captured using a Nikon D200 placed over the vase.…”

Downsampling scattering parameters for rendering anisotropic media

“…Although these approaches could be adapted to handle some scattering parameters, they normally offer limited accuracy. In addition, a number of techniques have been introduced to efficiently handle surface-based reflectance profiles (e.g., [Tan et al 2005;Bruneton and Neyret 2012;Jarabo et al 2014]), micro-geometry (e.g., [Westin et al 1992;Cook et al 2007;Jakob et al 2014]), or specialized volumes [Meng et al 2015]. Unfortunately, these methods are not easily applicable to our problem.…”