An Analytical Method for Rendering Plenoptic Cameras 2.0 on 3D Multi-layer Displays

Armand Losfeld, Nicolas Seznec, Laurie Van Bogaert, Gauthier Lafruit, Mehrdad Teratani

Abstract: Plenoptic cameras capture 3D information from the scene in the form of a dense light field thanks to an array of micro-lenses placed in front of the sensor. These cameras have great potential for future 3D remote applications. Therefore, it is crucial to explore effective rendering methods for 3D displays using this data. Until now, only a single pipeline using external software, which converts the plenoptic image as a dense multi-view, was proposed, which makes it slow and prone to artifacts. In this paper, we consider using the plenoptic image as the input light field data for conventional rendering methods, instead of the traditional multi-view format. Hence, we modify the camera projection to a plenoptic one for 3D display image generation without the need for external software. Our method only takes a few seconds and works with any plenoptic camera model.

External IDs:dblp:conf/mmm/LosfeldSBLT25