Present generations of 3D displays including stereoscopic and autostereoscopic displays have very limited number of views, and thus limited parallax. In contrast, the emerging light field (LF) displays support hundred(s) of views with acceptable spatial resolution thereby enabling a more realistic representation of 3D scenes. This is accomplished for the price of high data throughput, complex data acquisition and a high demand of computational power. Thus, the optimization of the content representation is of crucial importance for the performance of the whole display system. In this paper, we discuss the requirements for LF based processing of 3D content for representation on the new generation of ultra-realistic LF displays. We analyze the overall processing chain from acquisition through LF based modeling and representation to visualization on the considered displays. By analyzing the visualization capabilities of a given LF display using spatial and frequency domain analysis, we draw guidelines on how to properly acquire the required data and repurpose it based on the targeted display. We show that by taking into account the properties of the display during scene sensing and during LF processing, a good visual representation of 3D content on a given display can be achieved with a minimalistic capture setup.

Date Published: 25 September 2014
PDF: 15 pages
Proc. SPIE 9117, Three-Dimensional Imaging, Visualization, and Display 2014, 911710 (25 September 2014); doi: 10.1117/12.2062010

Published in SPIE Proceedings Vol. 9117:
Three-Dimensional Imaging, Visualization, and Display 2014
Bahram Javidi; Jung-Young Son; Osamu Matoba; Manuel Martínez-Corral; Adrian Stern, Editor(s)