HDR and wide gamut appearance-based color encoding and its quantification

Color coding efficacy is critical to the success of any image compression method. The initial quantization step that happens during encoding limits the accuracy and directly affects compression performance. If too many code words are spent in areas of the color space that make little difference to perception, compression is taxed. If other areas are under-sampled, contour (banding) and color shifting artifacts can result. Legacy color encodings such as 24-bit sRGB do not easily extend to wide color gamuts and high dynamic range due to perceptual non-uniformity issues. In this paper, we compare a number of different color encodings and bit depths for suitability in HDR and wide gamut applications using a new metric for evaluating encoding efficacy, Number of Distinguishable Colors.