Supported by the FWO research community Audiovisual Systems
|
|
Chair Philip Dutré Toon Lenaerts (Computer Graphics, K.U.Leuven) Bert De Decker (EDM, Universiteit Hasselt) Yannick Francken (EDM, Universiteit Hasselt) [Slides] Peter Vangorp (Computer Graphics, K.U.Leuven) [Slides] [More info] 10.50 - 11.30: Break |
11.30- 12.30 Session 2|
Chair Tom Mertens Cosmin Ancuti (EDM, Universiteit Hasselt) Dan Cernea (ETRO, VUB) Pieter Peers (ICT, University Southern California) [More info] 12.30 - 13.30: Lunch |
13.30 - 15.00 Special Session|
Chair Philippe Bekaert Eric Lafortune (Luciad) Marcus Magnor (Technische Universität Braunschweig) 15.00 - 15.45: Break |
15.45 - 16.45 Session 3|
Chair Ares Lagae Muath Sabha (Computer Graphics, K.U.Leuven) [Slides] Jurgen Laurijssen (Computer Graphics, K.U.Leuven) Tom Mertens (EDM, Universiteit Hasselt) [Slides] 18.30: dinner at Blue Olive in Hasselt |
Guest Speaker Marcus Magnor|
"Visual Computing between Physics and Perception" Photos and videos are very efficient means to gather information about the real world. To interpret real-world image data, typically the well-known constraints of physical laws are enforced to build up physically correct digital models of natural phenomena. By determining model parameter values of physics-consistent models, meaningful physical measurements may be obtained from image data. Once physics-compliant models are available, physically correct rendering becomes possible by faithfully simulating the laws of optics. Not just photo-, but physically-realistic rendering results are achievable. For many applications, however, renderings need not be physically realistic to be perceived as being authentic. Because images become meaningful only via interpretation by our visual brain, there exist many, as of yet uncharted, degrees of freedom for rendering beyond the unrelenting rigor of physics. This opens up interesting new opportunities for computer graphics and virtual reality to create visual realism from real-world footage without the need to comply to physical laws. In my talk, I will discuss and exemplify different approaches on how to recover digital models of real-world entities from image data. These models may be derived either by enforcing the (known) physics of the scene, or by regarding perceptional consequences only. While the former methods yield physically meaningful information about the scene, approaches of the latter kind allow for much easier creation of authentically appearing rendering results. More infomation about Marcus Magnor can be found at his home page. |
Venue
Registration