Participants in the video communications industry like to use the term “Quality of Experience,” but what does that term mean? Whose experience are they talking about?
Is it that of the engineers who monitor the video delivery chain? Is it the executives and content makers who decide what to deliver?
Or is it that of the end-user, the person sitting on the couch in their living room?
SSIMWAVE believes QoE starts and ends with the latter. After all, it’s for their benefit that a video exists and it’s their satisfaction that ultimately drives the entire industry.
A true QoE metric must, therefore, reflect the experience of end-users. It must “see” and predict video quality the way that the human visual system “sees.”
Many intermediate variables that are measurable at different points along the video delivery chain can influence the end-users’ QoE, but they should not be counted as a true QoE metric unless their impact is properly translated to a quantitative measure on how end users’ visual experience is affected – a score, in other words.
Moreover, any automated system aiming to measure true QoE must perceive what the end users perceive and say what end users say about the quality of their experience. For example, a true QoE metric should tell if a video is of excellent, good, fair, poor or bad quality, just like a human.
Video quality measures that do not take into account perceptual differences due to viewing conditions, such as viewer device type and size, video frame rate, and pixel resolution versus device viewing resolution, are not true QoE metrics. Video stream parameter and device playback behavior based measures, such as statistics of the average receiving bitrate and the duration/frequency of video freezing events, are merely QoS metrics at the device, or at best pseudo-QoE metrics, because they do not “see” the actual pixels of each video frame like humans and are not capable of capturing the large impact of both content variations and compression artifacts on visual quality.
Quality of experience is about the quality of the viewers’ experience. Viewers decide what’s good enough — or not.
The original article was published on Medium.com. Jan 4, 2019.