COMPUTATIONAL HOLOGRAPHY

RIP holograms: We're currently experimenting with a new method of generating holographic stereograms that combines light field rendering and physically based fringe computation. The resulting "RIP" holograms can be configured to meet the sampling requirements for a given light field, and they don't exhibit many of the aliasing and diffraction artifacts that are present in conventional computed holographic stereograms. The technique can exploit coherence among parallax views, is suitable for hardware implementation, and seeks to improve image quality without sacrificing computational speed. Software subsystems developed in C++ and openGL. This work is currently being prepared for publication with collaborators and co-authors Michael Halle and Ravikanth Pappu.

COMPUTATIONAL HOLOGRAPHY

Incremental computing: This technique for locally and incrementally computing interference modeled holograms was initially developed to rapidly update haptic holograms on the MIT MarkII holovideo display. The method offers most economy when only part of a scene changes at each simulation timestep. Incremental computing uses a precomputed table of fringes and performs simple operations on a "state" hologram as the scene geometry changes. Ultimately the frame rate is dependent on the number of hologram primtives affected by scene changes and on the number of HPO hologram lines they populate. Consequently, the technique is most appropriate for operating on wireframe or other sparsely populated scenes. For changes involving ~700 primitives, a 36MB hologram can be recomputed and updated on the display at 0.3 frames/second, a current best for MIT's holovideo. Software subsystems developed in C and C++ on Unix and Windows/PC platforms, and on the custom Holo-Cheops Image Processing System.