Computer graphics: Less computing time for sand

September 4, 2015
A digital sandcastle consists of millions of grains. Its photorealistic presentation by a computer now becomes more computation-efficient. Credit: KIT / Disney Research

Computer graphics today can produce amazingly photorealistic images. Many motives, however, require very long computation times. Researchers of KIT, Disney Research, Zurich, and Cornell University have now developed a process, by means of which granular objects made of e.g. sand, snow or sugar can be computed more quickly. It was presented recently at the renowned ACM SIGGRAPH 2015 International Conference for Computer Graphics in Los Angeles.

"Objects of granular media, such as a sandcastle, consist of millions or billions of grains. The computation time needed to produce photorealistic images amounts to hundreds to thousands of processor hours," Professor Carsten Dachsbacher of the Institute for Visualization and Data Analysis of KIT explains. Materials, such as sand, salt or sugar, consist of randomly oriented grains that are visible at a closer look only. Image synthesis, the so-called rendering, is very difficult, as the paths of millions of light rays through the grains have to be simulated. "In addition, complex scattering properties of the individual grains and arrangement of the grains in a system can prevent classical acceleration techniques from being used. This makes it difficult to find efficient algorithms," doctoral student Johannes Meng adds. "In case of transparent grains and long light paths, computation time increases disproportionately."

For image synthesis, the researchers developed a new multi-scale process that adapts simulation to the structure of light transport in on various scales. On the finest scale, when only few grains are imaged, geometry, size, and material properties of individual discernable grains as well as their packing density are considered. As in classical approaches, are traced through the virtual grains, which is referred to as path tracing. Path tracing computes light paths from each pixel back to the light sources. This approach, however, cannot be applied to millions or billions of grains.

The new process automatically changes to another rendering technique, i.e. volumetric path tracing, after a few interactions, such as reflections on grains, provided that the contributions of individual interactions can no longer be distinguished. The researchers demonstrated that this method normally applied to the calculation of light scattering in materials, such as clouds or fog, can accurately represent and more efficiently compute light transport in granular materials on these scales.

On larger scales, a diffusion approximation can be applied to produce an analytical and efficient solution for remaining transport. This enables efficient computation of photorealistic representation in case of bright and strongly reflecting grains, such as snow or sugar.

The researchers also succeeded in demonstrating how the individual techniques have to be combined to produce consistent visual results on all scales - from individual grains to objects made of billions of - in images and animations. Depending on the material, the hybrid approach accelerates computation by a factor of ten up to several hundreds compared to conventional path tracing, while image quality remains the same.

Explore further: New method efficiently renders granular materials at multiple scales

More information: Link to the publication on computer graphics, visual computing, and geometry processing by the KIT Institute for Visualization and Data Analysis: cg.ivd.kit.edu/granular_media.php 

Related Stories

Hydrodynamics approaches to granular matter

March 26, 2015

Sand, rocks, grains, salt or sugar are what physicists call granular media. A better understanding of granular media is important - particularly when mixed with water and air, as it forms the foundations of houses and off-shore ...

Scientist develops disposable devices to identify allergies

February 9, 2015

Gabriel Caballero Robledo from the Center for Research and Advanced Studies (CINVESTAV)at Monterrey, Mexico, is working on the design of a small medical device capable of detecting allergies or diseases quickly and at a low ...

Recommended for you

Volvo to supply Uber with self-driving cars (Update)

November 20, 2017

Swedish carmaker Volvo Cars said Monday it has signed an agreement to supply "tens of thousands" of self-driving cars to Uber, as the ride-sharing company battles a number of different controversies.

0 comments

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.