On The Implication Of Light Field Compression On Post-Processing Algorithms

Abstract

A Light Field (LF) can be parameterized as an assembly of beams in empty space (e.g. direction and intersection with plane with no starting point). As LFs are a 4D representation of a scene including both color intensity and angular information unlike the conventional 2D images which contain only color information, it is essential to adapt 4D LFs to be handled by file formats designed for assemblies of 2D images. This paper proposes adapting LFs to current production workflows, allowing interoperability between imaging applications and devices. OpenEXR is a standard that provides substantial advantages over conventional multiplexed video files or images, for special effects or in media environment. It can store many data channels, which allows for maximum flexibility in compositing. We have investigated the possibilities to store LFs in this format and also evaluated the different EXR compressions (lossy and lossless). In addition to this, a standard format such as JPEG XT which offers support for high bit depth image representation is also evaluated and compared.