Computational Imaging: The Race Against Time

Paul Debevec
University of Southern California, ICT

Slides (pdf)

Computational Imaging involves the measurement, interpolation, and extrapolation of many dimensions of photometric data -- most commonly, the position and directions of rays of incident and radiant light in a static scene. Recently, sophisticated devices have allowed capturing datasets such as light fields in time as well, making it possible to record dynamic scenes. In this talk I will present results in capturing time-varying reflectance fields that allow dynamic events to be visualized under novel lighting conditions. Such real-time techniques face additional acquisition challenges, as the basis lighting conditions must be rapidly time-multiplexed, data must be acquired and stored quickly, and scene motion must be compensated for. Furthermore, the amount of light appropriate for a living subject and the photon noise characteristics of sensors place limits on the amount of reflectance information that can be acquired. In this context I will discuss how reflectance models can be used to extrapolate information within such datasets, the extent to which exploiting reciprocity may help, and how time-resolved imaging may yield new techniques for analyzing object reflectance.


Paul Debevec is a research assistant professor in the Computer Science Department of the University of Southern California (USC) and executive producer of graphics research at the USC Institute for Creative Technologies. Debevec's Ph.D. thesis (UC-Berkeley, 1996) presented Facade, an image-based modeling and rendering system for creating photoreal architectural models from photographs. Using Facade he led the creation of virtual cinematography of the Berkeley campus for his 1997 film "The Campanile Movie" whose techniques were later used to create virtual backgrounds in the movie "The Matrix". Debevec subsequently developed techniques for capturing real-world illumination and illuminating synthetic objects with real light, demonstrating image-based lighting results in his films "Fiat Lux" (1999) and "The Parthenon" (2004). Debevec has also led the development of a series of Light Stage devices that allow objects, actors, and performances to be synthetically illuminated with novel lighting, recently used to create photoreal digital actors for the 2004 film "Spider Man 2". In 2001 Debevec received ACM SIGGRAPH's first Significant New Researcher Award, and in 2005 received a Lillian Gilbreth Lectureship from the National Academy of Engineering.