Originally published by ipfray.com: https://ipfray.com/part-2-the-value-of-modern-video-codec-technology-in-video-streaming/

The four modern video codecs licensed in Access Advance’s Video Distribution Patent Pool (VDP Pool) enable delivery of high-quality video and enhanced viewer satisfaction over real-world internet connections

One central challenge for every video service provider (VSP) is delivering the highest quality video at the best possible resolution over real-world network connections. Whether a platform relies on subscriptions, advertising, live commerce, or a hybrid model, revenue depends on keeping users engaged. Users disengage, reducing revenue across all models, when the video quality deteriorates.

This paper summarizes some key technical benefits of H.265/HEVC, H.266/VVC, AV1, and VP9 (the “modern video codecs” (MVCs)) that are due to their greater compression efficiency and that result in longer watch times, tighter engagement, and increased revenue across all video streaming business models.[1]

The MVCs enable streaming higher-quality video in all resolutions at lower bitrates solving for today’s crowded network connections

High-quality video typically includes (1) high resolution, such as 4K UHD video, which allows for clearer videos, (2) high dynamic range (HDR), which allows for more contrast, more brightness, and a broader palette of available colors, and (3) high frame rate (HFR) – more frames per second – which reduces blur and lag.

Netflix created the “video multimethod assessment fusion” (VMAF) metric, which is widely used to measure video quality as perceived by viewers. A VMAF score of 70 is the dividing line between “fair” and “good” video quality.

An uncompressed video file in 4K contains a huge amount of data and is even larger with HFR or HDR. The ability to transmit the file depends on the amount that data can be compressed while retaining video quality. On average, due to superior compression efficiency, 4K videos encoded in MVCs are more than 60% smaller than comparable videos encoded in the prior-generation standard H.264/AVC. Given typical network constraints on streaming such video, only MVCs can deliver “good” or better quality 4K video at the typical bitrates of today’s connections. AVC struggles to deliver 4K video at a “good” VMAF quality score over typical network conditions and cannot reliably deliver good-quality UHD video when network conditions deteriorate (e.g., either due to low bandwidth capacity or heavy traffic).

The MVCs also improve the quality of the user’s video experience at resolutions below UHD including for HD and SD video. Their superior compression efficiency reduces the bitrate required to stream HD content by 42 percent to 49 percent, and 540p resolution content by 23 percent to 34 percent. Those reductions in network bitrate requirement mean that at any given network bandwidth, MVCs deliver a higher-quality video experience compared to AVC, resulting in increased watch time and engagement for a given video, which increases VSP revenue under every business model.

MVCs reduce churn by improving rebuffering, bitrate switching, and startup times

The MVCs licensed in the VDP Pool minimize rebuffering, reduce bitrate switching, shorten startup times, and maintain video quality during periods of network congestion. The resulting better viewing experience increases watch time, reduces customer churn, and strengthens user satisfaction, resulting in higher revenues for VSPs across all business models.

            MVCs reduce rebuffering

When a VSP begins streaming a video, it transmits to the viewer both the initial portion of video it will play as well as an additional portion that can be played if the network transmission is interrupted or slowed. That additional portion is termed a “buffer”. “Re-buffering” is the condition that occurs if the media playback has exhausted the buffer and must be paused while more data is downloaded. Microsoft measures the degree of rebuffering by calculating a “rebuffering ratio” defined as the buffering time as a share of the total of playback plus buffering time. Microsoft characterizes a rebuffering ratio of less than 1% as a good experience, and a ratio of more than 3% as a poor experience. Similarly, Amazon’s CloudFront content distribution network (CDN) displays the rebuffering ratio on its Quality of Experience (QoE) dashboard. BlazingCDN, identifies its improvement in rebuffering ratio as a measure of quality directly affecting users.

Higher rebuffering ratios reduce watch time, that is the time the viewer spends watching the video. Industry studies have found:

• the rebuffering ratio was the most important factor in determining watch time
• as the rebuffering ratio increases, the time spent watching a video stream decreases
• a single rebuffering event led to a 39-percent reduction in watch time for streamed videos
• 4 rebuffering events led to a 48-percent reduction in watch time.
• a one-percentage-point increase in the rebuffering ratio over the course of a 90-minute live event reduced watch time by three minutes

Studies have also confirmed that watch times increase consistently and significantly with reduction in the rebuffering ratio that can be obtained from using MVCs to stream HD video.

MVCs reduce bitrate switching

VSPs typically employ adaptive bitrate (ABR) streaming, encoding video segments in different qualities, which in turn require different minimum network bitrates/bandwidth to distribute, and combine them into an ABR “ladder” to distribute the video. ABR streaming sends the viewer the segment on the ladder in the highest possible bitrate based on the conditions of the network and the viewer’s device at any given time, enabling viewers to continue watching a video as network conditions vary.

A change or “switch” in bitrate – such as to a lower bitrate when network congestion increases – can reduce the need for additional rebuffering. But the tradeoff is that the viewer might also experience a noticeable drop in video quality, as the frequency of bitrate switching increases with deteriorating network quality. Therefore, VSPs typically try to avoid bitrate switches.

One study found that 90 percent of YouTube users abandoned a session within 60 seconds of a single bitrate reduction. Another study observed that an 88.8-percent reduction in bitrate switching increased mean watch time by 5.91 percent.

Due to their superior compression efficiency MVCs reduce the bitrate necessary to stream video on every step of the ABR ladder, reducing the need for and frequency of, bitrate switching. Netflix reported that the switch from one MVC to another—from VP9 to AV1—resulted in a 38-percent decrease in the rate of noticeable quality reductions caused by bitrate switching while streaming Netflix, due to AV1’s 20-percent greater compression efficiency relative to VP9. Moving from AVC to MVCs would produce an even larger decrease in bitrate switching given that the average compression-efficiency improvement moving from AVC to MVCs is twice that of moving from VP9 to AV1.

            MVCs Decrease Startup Time

By reducing the size of both the initial portion of video to be played as well as a portion that represents an initial buffer, MVCs reduce the “startup time” for the streamed video, the interval between when the user clicks on the link for a video and when the video begins playing. Longer startup times reduce user engagement and watch time.

Several industry studies have shown that startup time is a significant factor in viewer disengagement and that MVCs reduce startup time:

• a 50-percent decrease in required bitrate reduced the startup time for streamed videos by approximately 10 percent (Netflix)
• moving from AVC to VP9 resulted in a 15% to 80% faster startup time for streamed videos due to the superior compression efficiency
• a one-second increase in startup time results in a 5.8-percent increase in the rate at which viewers abandon the video

It is obvious that startup time delays are a significant problem for short-form video, such as on social media; for a video that lasts less than 10 seconds, for example, even a few seconds of startup time causes viewers to abandon the video. But the effect is not limited to short-form video. The third bullet-point above is sourced from a well-known U. Mass/Akamai study.[2] That study included both longer form videos, defined in the study as 30 minutes or longer, as well as shorter videos, and the 5.8% abandonment rate is an average across video types. At the time of the study, long-form video predominated and short-form video, while well known today, was just getting off the ground, meaning the study is more a study of startup time in long-form than in short-form video. Thus, the U.Mass/Akamai study makes it clear that startup time is an issue in long-form videos as well as in short-form video, and the benefits of MVCs in reducing startup time accrue to both short and long video, increasing viewer engagement and watch times—and therefore revenue—across all VSP business models.

Access Advance’s VDP Pool is the right decision – the smart technical and business decision – for video providers across all business models

The technical features of MVCs provide the same benefits across all business models, even though different business models monetize those benefits differently.  Subscription video-on-demand (SVOD) services benefit from higher customer retention and greater lifetime value. Free ad-supported television (FAST) services and social media platforms benefit from longer viewing sessions, more advertising inventory, and stronger engagement. Live commerce platforms benefit from higher viewer retention and increased conversion rates.

For VSPs making long-term licensing decisions, the Access Advance VDP Pool provides access to the best technology for video streaming in an innovative, predictable, flexible, and cost-effective pool license. Its growing industry adoption—including licensees with major global platforms such as Meta, Alibaba, ByteDance, Kuaishou, Tencent, NTT Docomo, Roku, and Youku, as well as the 46 licensors who have already joined the pool—demonstrates increasing market confidence in the pool’s value proposition as the right pool for video streaming.

---

[1] This paper is based on the Criterion study “The Economics of Video Compression: Why the Access Advance Video Distribution Patent Pool Is Fair, Reasonable, and Nondiscriminatory” which can be requested at https://accessadvance.com/policy-advocacy/the-economics-of-video-compression-why-the-access-advance-video-distribution-patent-pool-is-fair-reasonable-and-nondiscriminatory/

[2] Krishnan, S. S., & Sitaraman, R. K. (2012). Video stream quality impacts viewer behavior: Inferring causality using quasi-experimental designs. In Proceedings of the 2012 ACM Conference on Internet Measurement Conference (pp. 211-224). doi.org; found at https://dl.acm.org/doi/10.1145/2398776.2398799?ref=swarmify.com