15 maggio 2012

New Results for Disney Research at Eurographics 2012, Cagliari (Sardinia Island)

 The Disney Research, Zurich (DRZ) graphics and materials research teams, along with collaborators from Swiss Federal Institute of Technology (ETH Zurich), Massachusetts Institute of Technology (MIT), Technische Universität Berlin (TU Berlin), Max-Planck-Institut Informatik (MPI Informatik Saarbrucken, Germany), and Disney Research, Boston (DRB) have five papers accepted for presentation at the 2012 Eurographics Conference to be held May 13-18 in Cagliari, Sardinia (Italy).

“Disney Research is pleased to return to Eurographics and to have strong representation with this year’s selected works. The research displayed at this year’s conference reinforces The Walt Disney Company’s strong commitment to technology and innovation in entertainment” states Markus Gross, Director, Disney Research, Zurich.

The presentations to be shared at the conference include a physical object whose surface can display up to four distinct images depending on the direction of the illumination, and a method for designing physical objects that cast shadows of different shapes and colors when illuminated from different angles. Additional presented research focuses on image rendering and video processing, on physical simulation of cloth using measurements from real life examples, and on an automatic process for creating custom shaped rubber balloons.

Project abstracts can be found below

SHADOWPIX (Multiple Images from Self Shadowing) are white surfaces that display several prescribed images formed by the self-shadowing of the surface when lit from directions. The effect is surprising and not certain commonly seen in the real world. We present algorithms for constructing SHADOWPIX that allow up to four images to be embedded in a single surface. SHADOWPIX can produce a variety of unusual effects depending on the embedded images: moving the light can animate or relight the object in the image, or three colored lights may be used to produce a single colored image. SHADOWPIX are easy to manufacture using a 3D printer.
Authors: Amit Bermano (DRZ, ETH Zurich), Ilya Baran (DRZ), Marc Alexa (TU Berlin, Disney Research Boston), Wojciech Matusik (DRZ, MIT CSAIL, DRB)

Manufacturing Layered Attenuators for Multiple Prescribed Shadow Images. We present a practical and physical objects that cast different color shadow images when lit from inexpensive method for creating different directions. Given target images and light directions, this approach computes attenuation masks, which are then printed on transparent materials and stacked to form a single layered attenuator whose shadows look like the input images. Alternatively, this method can compute layers so that their permutations produce different prescribed shadow images under fixed lighting.
Each layered attenuator is quick and inexpensive to produce, can generate multiple full-color shadows, and can be designed to respond to different types of natural or synthetic lighting setups. The effectiveness of the layered attenuators is illustrated in simulation and in reality with the sun as a light source.
Authors: Ilya Baran (DRZ), Philipp Keller (ETH Zurich), Derek Bradley, Stelian Coros, Wojciech Jarosz and Derek Nowrouzezahrai ( DRZ), Markus Gross (DRZ and ETH Zurich).

NoRM: No-Reference Image Quality Metric for Realistic Image Synthesis. Images and video frames of complex 3D scenes using photo-realistic rendering software often contain perceptually significant artifacts. Expert parameter tuning is required to avoid these artifacts. Detecting and preventing artifacts can be automated through quality evaluation of images, thus avoiding a tedious manual process. Most practical quality assessment methods rely on a ground-truth reference, which is often not available in rendering applications. We show that the performance of a dedicated algorithm without a reference metric can match the state-of-the-art algorithms that do require a reference. This method successfully detects various non-trivial types of artifacts, such as VPL clamping, glossy VPL noise and shadow map aliasing. In a subjective study we show that, on average, the correlation between our metric's predictions and the ground truth data is slightly higher than the full-reference metrics HDR-VDP-2 and SSIM.
Authors: Robert Herzog and Martin Cadík ( MPI Informatik Saarbrucken, Germany), Tunç O. Aydın (MPI Informatik Saarbrucken, Germany, DRZ ), Kwang In Kim, Karol Myszkowski and Hans-Peter Seidel (DRZ)

Data-Driven Estimation of Cloth Simulation Models. Modern cloth simulators produce beautiful results, but an accurate match to the
behavior of a real piece of cloth is often difficult to achieve. This paper provides
measurement and fitting methods that allow nonlinear models to be fit to the observed deformation of a real cloth sample. Unlike standard textile testing, our system measures complex 3D deformations of a sheet of cloth, not just one-dimensional force-displacement curves, so it provides a good match for a wider range of deformation conditions. The fitted models are evaluated by comparison to measured deformations with motions very different from those used for fitting.
Authors: Eder Miguel (URJC Madrid, DRZ), Derek Bradley, Bernhard Thomaszewski and Bernd Bickel (DRZ), Wojciech Matusik (DRZ, MIT), Miguel A. Otaduy (DRZ), Steve Marschner (Cornell University)

Computational Design of Rubber Balloons.  These balloon animals are not your standard carnival fare! We present an automatic process for fabrication-oriented design of custom-shaped rubber balloons. Given a detailed target shape, we compute an optimal balloon when inflated, approximates the target shape as shape that, closely as possible.
To achieve this goal, we propose a novel physics-driven shape optimization method which combines physical simulation of inflatable elastic membranes with a dedicated constrained optimization algorithm. By fabricating balloons designed with our method, and comparing their inflated shapes to the results predicted by simulation, we are able to validate our approach. An extensive set of manufactured sample balloons demonstrates the shape diversity that can be achieved by our method.
Authors: Mélina Skouras (ETH Zurich), Bernhard Thomaszewski and Bernd Bickel (DRZ), Markus Gross (ETH Zurich, DRZ)

Disney Research is a network of research laboratories supporting The Walt Disney Company. Its purpose is to pursue scientific, and technological innovation to advance the Company's broad media and entertainment efforts. Disney Research is managed by an internal Disney Research Council co-chaired by Disney-Pixar's Ed Catmull and Walt Disney Imagineering's Bruce Vaughn, and including the directors of the individual labs. It has facilities in Los Angeles, San Francisco, Pittsburgh, Boston, and Zurich. Research topics include Computer Graphics, Video Processing, Computer Vision, Robotics, Radio and Antennas, Wireless Communications, Human-Computer Interaction, Displays, Data Mining, Machine Learning and Behavioral Sciences.

Eurographics 2012, the 33rd Annual Conference of the European Association for Computer Graphics, will take place from May 13 through May 18 in Cagliari, Sardinia, Italy. It will be organized by CRS4 Virtual Computing and University of Cagliari.

Full conference participation downloadable information on Eurographics 2012.

Nessun commento: