 |
|
|
16. Video-Based Rendering
Half-Day, Monday, 1 August, 8:30 am - 12:15 pm
Room 515A
Level: Intermediate
A thorough introduction to how to acquire and process multiple video streams for omni-perspective, interactive rendering of real-world, dynamic scenes. Attendees learn how to reconstruct and represent dynamic scene geometry from multi-video footage, as well as how to render time-varying scenes video-realistically from arbitrary viewpoints in real time.
Prerequisites
Familiarity with the concept of image-based rendering. Some basic knowledge of computer vision fundamentals is helpful but not mandatory.
Intended Audience
Graduate students interested in interdisciplinary graphics and vision research, and professionals from the movie, special effects, and interactive entertainment industries.
Co-Organizers
Marcus Magnor
MPI Informatik
Marc Pollefeys
University of North Carolina at Chapel Hill
Lecturers
German Cheung
Neven Vision Inc. (formerly Carnegie Mellon University)
Wojciech Matusik
Mitsubishi Electric Research Laboratory (MERL)
Christian Theobalt
MPI Informatik
Schedule
| 8:30 |
Introduction
Magnor
|
| 8:45 |
Multi-Video Acquisition
Cheung
|
| 9:05 |
Small-baseline VBR
Matusik and Magnor
|
| 9:45 |
Wide-baseline VBR
Matusik and Pollefeys
|
| 10:15 |
Break
|
| 10:30 |
Camera Calibration
Pollefeys
|
| 11 |
Spacetime Coherence
Cheung and Magnor
|
| 11:30 |
Model-Based VBR
Cheung and Magnor
|
| Noon |
Outlook & Discussion
Pollefeys
|
|
|
|
|
