The Computer Graphics Forum 2017 Cover Image has been selected by the CGF editorial board. We thanks all people who submitted this year, and hope that they will participate to the next year contest.

Winner of the Computer Graphics Forum 2017 Cover Image Contest

Alessia Marra (1), Maurizio Nitti (1), Marios Papas (1), Thomas Müller (2), Markus Gross (2), Wojciech Jarosz (3) and Jan Novák (1)

(1) Disney Research
(2) Disney Research, ETH Zürich
(3) Dartmouth College

This is a frame of an animation sequence where white sugar, brown sugar, and cinnamon are mixed inside a wooden bowl (full animation: ). Our algorithm accurately preserves the important global illumination grain appearance such as glints, translucency, and occlusion in this scene. The grain placement was simulated in Houdini and given as input to our method which handles rendering and hallucinates temporally consistent grain orientations.

The key to our efficiency lies in two complementary grain scattering aggregation techniques which are automatically chosen to minimize approximation error, without any input from the user. Our method can accelerate light transport in dynamic, spatially varying, and polydisperse granular media. Grain placement can be modeled procedurally or with off-the shelf particle simulation tools.

Project Page:

Second place:

Tobias Günther (1), Alexander Kuhn (2), Hans-Christian Hege (2), Markus Gross (1) and Holger Theisel (3)

(1) ETH Zürich
(2) Zuse Institute Berlin
(3) University of Magdeburg

This visualization shows our unbiased Monte Carlo rendering of atmospheric air flow patterns using the finite-time Lyapunov exponent (FTLE) field. Our method was published at the Eurographics Conference on Visualization 2016.

The example shows atmospheric structures in a reanalysis simulation of the European Centre for Medium-Range Weather Forecasts (ECMWF). The FTLE field highlights areas of strong repelling behavior and denotes regions that cannot be crossed by virtual atmospheric tracers. These so-called material structures govern, for instance, the advection of tracers (like dust or carbon dioxide), temperature diffusion, and cloud formation.

Our rendering approach enables high-quality volumetric visualizations of complex FTLE fields and is guaranteed to faithfully represent detailed ridge surface structures. While previous methods suffered from grid discretization and ray marching artifacts, we treat the volume as participating media and employ a consistent light transport simulation method, based on unbiased free path sampling.

Third place:

Eric Guérin, Eric Galin, Adrien Peytavie, Francois Grosbellet and Jean-David Genevaux


The image shows two statues featuring tens of thousands of small entangled objects (grass tufts, branches and rocks). Our method consists in computing a generic structure, the Ghost Tile, that stores a dense collection of partially overlapping candidate geometric objects in a tile and
a collision graph between those objects. This structure is then combined and processed with a user- or procedurally-defined control volume (here the statue) and density functions to generate
a distribution of different non-intersecting entangled instances constrained to the control volume. Our method can generate hundreds of thousands of densely packed objects of different type interactively (in this case approximatively 20k instances were instantiated in less than a second). Mitsuba ray-tracer was used to render the scene.

The method and further details may be found in "Efficient modeling of entangled details for natural scenes", Computer Graphics Forum, Volume 35, number 7, 2016.