Traditional digital imaging techniquesattempting to capture scenes with wide ranges of lightingsuffer areas of under-exposed or over-exposed pixels. HighDynamic Range (HDR) techniques, however, are anexception, being especially designed to capture the full rangeof lighting in a real scene, including even more than thehuman eye can see. To handle the extra data required withHDR requires effective compression. One approach gainingpopularity is to consider some form of one-stream transferfunction to map the HDR content to a fixed number of bits,typically 10 or 12. While it is true that many scenes do containa dynamic range of lighting that is sufficiently low to beadequately contained within 12 bits, there are many othersfor which 12 bits are not enough. This paper demonstrateshow significant information present in the original scene, andthe flexibility to display it in a manner best suited to theviewer, can be lost with a one-stream approach to HDR videoencoding.
展开▼