Thursday December 2 , 10h30 - 11h30 , room F107 , Seminar

Chrstian Theobalt (MPI Saarbrücken )

Model-based Performance Capture for High-Quality Dynamic Scene Reconstruction from Multi-view Video

Convincing human characters are an essential component of most simulated virtual environments. In this talk, I will present some of the algorithms we developed at MPI Informatik to capture high-quality dynamic animation models of human actors from multi-view video data of real individuals. Our algorithms enable us to capture dynamic shape, appearance and motion of people without optical markers. In particular, I will present a tracking method that is based on hierarchical mesh deformation, as well as a hybrid capturing approach that is based on a combination of a skeleton and a deformable surface model. With our algorithms, we are able to reconstruct detailed dynamic scene models even of people wearing loose apparel, such as a dress.

I will also show how we can reconstruct editable models of human actors from multi-view video. To this end, we automatically separate the reconstructed performances into non-cloth and cloth regions, and estimate the parameters of a physics-based cloth model for the latter. This way, we are able to change the motion of a reconstructed actor arbitrarily during postprocessing, and still get animations with realistic cloth deformations. If time allows, I will also show how performance capture techniques can be used for more general 2D and 3D video editing operations, such as changing the body shape of an actor in a movie.

Biography: Christian Theobalt is the head of the research group "3D Video and Vision-based Graphics" at MPI Informatik. From 2007 until 2009 he was a visiting assistant professor in the Department of Computer Science at Stanford University. He received his MSc degree in Artificial Intelligence from the University of Edinburgh, Scotland, and his Diplom (MS) degree in Computer Science from Saarland University, Saarbruecken, Germany, in 2000 and 2001 respectively. From 2001 to 2005 he was a researcher and PhD candidate in Hans-Peter Seidel’s Computer Graphics Group at MPI Informatik. In 2005, he received his PhD (Dr.-Ing.) from Saarland University and MPI. His research interests include 3D video and dynamic scene reconstruction, marker-less optical motion capture, 3D computer vision, machine learning for graphics and vision, as well as image- and physics-based rendering. In 2007, he was awarded the Otto Hahn Medal of the Max-Planck Society, and in 2009 he received the EUROGRAPHICS Young Researcher Award.