Traditional computer graphics assumes carefully built content, namely accurate geometry, materials and lights; the difficulty and expense of creating such data limits the impact and widespread use of 3D computer graphics (CG). In contrast, advances in image- and video-based techniques allow easy creation and display of 3D content; these approaches are however very limited and not well integrated in traditional CG. In this talk, we propose some initial ideas on how to bridge the gap between these two -- often very separate -- worlds, leading to a unified approach to 3D CG. The key to our methodology is the interaction between traditional simulation-based forward CG and optimization/relaxation methodologies taken from inverse problem domains. We will illustrate these ideas with several examples of our recent work, where 2D image data guides simulation-based CG techniques, or where display and manipulation of incomplete or inaccurate 3D data is regularized with 2D image content. We show specific examples where applications such as simulation of weathering effects, texture synthesis, image-based rendering and image-based relighting can benefit from these ideas. The fundamental difference to traditional inverse problem methods is that our ultimate goal is the (forward) generation of plausible new image content. We will close our talk with thoughts on how to evaluate such plausibility with human perception and how to generalize -- and potentially formalize -- these ideas.
George Drettakis graduated from the Dept. of Computer Science of the University of Crete, Greece, where he participated in European projects at ICS/FORTH. He obtained an M.Sc.and a Ph.D., (1995) at the Dept. of C.S. (Dynamic Graphics Project) at the University of Toronto, Canada, under the supervision of Eugene Fiume. He was an ERCIM postdoctoral fellow Grenoble, Barcelona and Bonn. He started as a permanent Inria researcher in the iMAGIS group in Grenoble in 1995. He obtained his "Habilitation" at the University of Grenoble in 1999. In 2000 he moved to Inria Sophia-Antipolis, where he founded the REVES research group. He became a Inria Senior Researcher in 2003. He received the Eurographics (EG) Outstanding Technical Contributions award in 2007 and is an EG fellow. He is co-editor in chief of IEEE Trans. on Computer Graphics and Visualization and leads the EG working group on Rendering. He has worked on many different topics in computer graphics, with an emphasis on rendering. The first part of his career concentrated mainly on lighting and shadow computation. In the first years of the REVES group he worked on 3D audio and in particular perceptually-driven algorithms. In recent years he has worked on image-based rendering and relighting, textures, perception for graphics, weathering simulation,virtual reality and 3D interaction.