PRAHA CVUT
EG.ORG
tutorial 5
T5  New Trends in 3D Video

Date: Monday, September 3rd, Time: 09:00 - 12:30, Location: Lecture Room 309

Organizers:  Christian Theobalt (Stanford University, USA)
Stephan Würmlin (LiberoVision AG, Switzerland)
 
Speakers:  Christian Theobalt (Stanford University, USA)
Stephan Würmlin (LiberoVision AG, Switzerland)
Edilson de Aguiar (MPI Informatik, Germany)
Christoph Niederberger (LiberoVision AG, Switzerland)


Abstract

3D Video is an emerging and challenging research discipline that lives on the boundary between computer vision and computer graphics. The goal of researchers working in the field is to extract spatio-temporal models of dynamic scenes from multi-video footage in order to display them from user-controlled synthetic perspectives. 3D Video technology has the potential to lay the algorithmic foundations for a variety of intriguing new applications. This includes stunning novel visual effects for movies and computer games, as well as, facilitating the entire movie production pipeline by enabling virtual rearranging of cameras and lighting during post-processing. Furthermore, 3D Video processing will revolutionize visual media by enabling 3D TV and movies with interactive viewpoint control, or by enabling virtual y-arounds during sports-broadcasts. To achieve this purpose, several challenging problems from vision and graphics have to be solved simultaneously. The speakers in this course will explain the foundations of dynamic scene acquisition, dynamic scene reconstruction and dynamic scene rendering based on their own seminal work, as well as related approaches from the literature. They will explain in more detail three important categories of algorithms for dynamic shape and appearance reconstruction, namely silhouette-based, stereo-based, and model- based approaches. Alternative methods, such as data-driven approaches, will also be reviewed. The tutorial will focus on latest 3D Video techniques that were not yet covered in a tutorial, including algorithms for free-viewpoint video relighting, model-based deformable mesh tracking, as well as high-quality scene reconstruction with camera/projector setups. The course keeps a balance between the explanation of theoretical foundations, engineering problems and emerging applications of 3D Video technology. We therefore believe that the course will be a valuable and entertaining source of information for students, researchers and practitioners alike.


    



Extended Summary

  1. Introduction (15 min)

    Speaker: Christian Theobalt

    • 3D Video - Why bother?

  2. Silhouette-based Methods (25 min)

    Speaker: Stephan Würmlin

    • Silhouette-based Methods - Foundations
    • Point Primitives for 3D Video
    • Real-time Applications in tele-presence systems (the blue-c)

  3. Stereo-based Methods (25 min)

    Speaker: Stephan Würmlin

    • Stereo-based Methods - Foundations
    • Using Camera Systems and Structured Light for High-quality 3D Video
    • Postprocessing Methods

  4. Model-based 3D Video I (25 min)

    Speaker: Christian Theobalt

    • Foundations
    • Marker-less Tracking and Dynamic Scene Reconstruction
    • Model-based 3D Video Rendering

  5. Break

  6. Model-based 3D Video II (25 min)

    Speaker: Edilson de Aguiar

    • Alternative Model-based Approaches
    • Deformable Mesh Tracking for Relightable 3D Video

  7. Free-Viewpoint Video Relighting (25 min)

    Speaker: Christian Theobalt

    • Data-driven Dynamic Scene Relighting
    • Model-based Free-Viewpoint Video Relighting

  8. Applications (30 min)

    Speaker: Christoph Niederberger

    • Authoring and Editing 3D Video
    • Applications of 3D Video in Movie and TV Production

  9. Outlook And Discussion (10 min)

    Speaker: Stephan Würmlin

    • Conclusions
    • Questions


Speakers' Background

Christian Theobalt is a Postdoctoral Research Fellow in the Department of Computer Science at Stanford University. He is also the head of the research group "3D Video and Vision-based Graphics" in the Max-Planck-Center for Visual Computing and Communication (Saarbrücken/Stanford). Christian Theobalt received his MSc degree in Artificial Intelligence from the University of Edinburgh, Scotland, and his Diplom (MS) degree in Computer Science from Saarland University, Saarbrücken, 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. In 2006, he worked as a postdoctoral researcher in the Computer Graphics Group at MPI Informatik. His research interests include free-viewpoint and 3D video, marker-less optical motion capture, 3D computer vision, image-based rendering, computer animation and physically-based rendering.

Stephan Würmlin is a post-doctoral researcher in the Computer Graphics Laboratory at ETH Zurich and managing director of the blue-c-II project. He received a diploma degree in Computer Science from ETH Zurich in 2000 and a PhD degree from ETH Zurich in 2004. His PhD thesis focused on the design of the 3D video technology for the blue-c collaborative virtual reality system. His current research interests include 3D video, video-based rendering, and multimedia applications. Stephan Würmlin is co-founder and CEO of LiberoVision AG, a company focusing on commercialization of 3D video technologies for content enhancement of TV sports broadcasts.

Edilson de Aguiar received the BS in Computer Engineering from the Espirito Santo Federal University, Vitoria, Brazil, in 2002 and the MSc degree in Computer Science from the Saarland University, Saarbruecken, Germany, in 2004. He is currently working as a PhD student in the Computer Graphics group at the Max-Planck-Institut (MPI) Informatik, Saarbruecken, Germany. His research interests include 3D video, virtual and augmented reality, motion capture and computer animation.

Christoph Niederberger is founder and CTO of LiberoVision, a spin-off company of ETH Zurich (www.liberovision.com). He designed and implemented the prototype and leads the development of the LiberoVision products. He received a diploma degree in computer science engineering from ETH Zurich in 2001 and a PhD degree in computer graphics and artificial intelligence from ETH Zurich in 2005. In his PhD thesis, he investigated behavioral modeling of artificial agents for real-time environments such as computer games. After finishing his thesis, he became the chief software architect of the project NOVA at main station Zurich (nova.ethz.ch) where he was responsible for the design and implementation of the software for operations and content acquisition.



« Back
Last update : 12.05.2011

If you have any comments regarding this webpage please contact EG07 staff.