Perceptually-optimized content remapping for automultiscopic displays

3D content and display technology is now widely available. However, available displays range from large-screen cinematic projection systems to hand-held devices and from screens supporting glasses-free 3D modes to glasses-bound systems. Counterintuitively, the produced content is usually only generated for a single display configuration, making labor-intense, manual post-processing of the data necessary. Recently, several content remapping techniques for stereo content have been proposed (see e.g. [Lang et al. 2010]). We present a perceptually-driven optimization framework for automatic light field remapping, specially designed for automultiscopic displays. It takes into account both the limitations of the target display as well as those of a human observer, and poses the problem as a non-linear least squares optimization. Our model includes depth of field limitations and the contrast sensitivity function, as well as sensitivities to binocular disparity and motion parallax in the perception of depth. Additionally, the model can be adapted for more common stereo remapping.