Choosing Encoding Parameters for High-Dynamic Range Streaming
Sean T. McCarthy
This paper provides data and analysis on the impact of various high-dynamic range (HDR) encoding options on video quality and bitrate. Specifically, this paper examines permutations of perceptual quantizer (PQ), hybrid-log gamma (HLG), ${Y}^\prime ~C_{B}C_{R}$ , IC$_{T}C_{P}$ , encoded resolution, and bitrate using the open-source encoder $\times 265$ . Video distortion for each permutation is measured using publicly available nonproprietary objective video quality metrics. Today, we have two different ways of mapping light to HDR code values: PQ and HLG. We also have the option of representing color and luminance as either ${Y}^\prime ~C_{B}C_{R}$ or the newer alternative IC$_{T}C_{P}$ . For HDR adaptive bitrate streaming, we need to choose which compression levels and encoding resolutions to use when we create adaptation sets; but we do not yet have a complete understanding of how resolution and compression could interact to alter the highlights and deep darks that make HDR so visually potent. When viewed altogether, the many permutations of HDR encoding can make designing an HDR streaming service complicated and uncertain. This paper provides practical data and analysis that can be used to make designing HDR streaming services easier and more predictable.
This paper provides practical data and analysis that can be used to make designing HDR streaming services easier and more predictable.
Print ISSN
1545-0279
Electronic ISSN
2160-2492
Published
2018-04
Content type
Original Research
Keywords
High-dynamic range (HDR), high-efficiency video coding (HEVC), ICTCP, PQ. HLG, streaming, video quality, ×265, Y′CBCR