Netflix announced this week that it is deploying a version of HDR for its programming that optimizes video quality across all device categories ranging from TVs to mobiles and tablets.
HDR-VMAF, (Video Multimethod Assessment Fusion) which Netflix publicly announced in 2020, uses Netflix’s machine learning-based technology, dubbed “Dynamically Optimized (DO) encoding” to measure HDR quality by focusing on signal characteristics (as a result of lossy encoding) instead of display characteristics. Prior to that, the company was using a form of adaptive bitrate streaming that uses predetermined bitrates—regardless of content characteristics—for HDR video streaming.
Although the streamer first publicized VMAF in 2020, the pandemic delayed in-person testing of the format. As an alternative ,Netflix worked with Dolby Laboratories to conduct subjective tests with 4K-HDR content using high-end OLED panels in calibrated conditions created in participants’ homes. Netflix A/B tested HDR-DO encodes in production in the last half of 2021 and had optimized the company’s entire HDR catalog with VMAF by June 2023.
Neflix said deploying VMAF fully across its catalog is “reducing storage footprint and Internet data usage — and most importantly, improving the video quality for our members.”
