Quality Street

September 27, 2010
For an informed view on connected entertainment in the UK & Ireland, visit Cue Entertainment 

Although it might appear self-evident, a home entertainment delivery platform must place quality above all other factors particularly if the consumer is prepared to pay for the content shown on it.

Research to back up this point of view, however, is scarce, which is why “What matters in quality video online,” a report from industry analyst Screen Digest delivered this week by Head of Broadband Dan Cryan, is a welcome contribution to the discussion.

Among the data presented as part of an internet “webinar” was the forecast that annual long-form online TV consumption in the UK will reach 1 billion hours by 2014. For this is to be achieved it will be necessary to pin down the consumer perception of broadband video quality and, as far as possible, to meet user expectations. While some of the information was necessarily quite technical, everyone planning to deliver video online should heed the conclusion that consumers are willing to pay a premium for high definition (HD) broadband content.

Screen Digest joined forces with the internet backbone company Level 3 to analyse how users judge video quality. Level 3 is a B2B operation providing the high-speed connections that link internet service providers (ISPs) around the world to the internet. Without the fibre optic cables across continents and under the oceans that companies such as Level 3 provide, broadband networks would grind quickly to a halt.

Thanks to the intimate connection to video service operators in broadcasting, film and sports networks enjoyed by Level 3, key individuals at 17 of the largest suppliers of online video content, including Disney ABC, BBC, Sky and RTL Group, responded with their views of how consumers evaluate quality.

More than 70% of respondents cited high quality as the most important driver of online video consumption and over half of the remaining 30% said quality increases the willingness of users to pay for content. As Cryan puts it, “Online operators are either going to get more viewers or make more money by having a higher quality feed.”

The average bitrate of video from premium video sources in the UK is below 600 kilobits per second (kbps) currently, according to Screen Digest, between five and 10 times lower than the average DVD and far less than Blu-ray Disc. It is expected to rise to just over 800 kbps by 2014, still barely adequate for an HD service.

Screen Digest obtained BBC data for the iPlayer for the presentation, which revealed that there is a close relationship between viewing times and bit rates. “At 1.5 megabits per second (Mbps) the average viewer watches for 27 minutes,” says Cryan, “at 800 kbps that falls to 22 minutes and at the lowest bit rate of 480 kbps the average is 17 minutes.”

“The people we interviewed consistently reported that quality improves consumption,” Cryan says. “First, by making people engage with the brand more, so they return to the site and consume more video. Second, by reducing the number of viewers turning off when the quality isn’t there and third, by increasing viewing time for ‘second tier’ content.”

Users clearly make value judgements of what is good and bad when it comes to watching online TV and they choose their service accordingly. Operators need to know what attracts viewers and what keeps them watching, so what comes out as the most important factor in assessing quality – high definition; colour quality; frame rate?

In fact it is none of these things, at least according to the 17 senior individuals who were questioned. The clear winner in the “what defines quality?” stakes was “well synchronised sound.”

Sound that is closely tied into the action, especially lip-synch in music and drama, was said to be the most critical, with the “sweet spot” for an acceptable level of delay being 15-20 milliseconds (thousandths of a second) nominated by 60% of those questioned. Cryan points to the “thwack of the racket” in tennis as an example of the importance of synch in sport though the respondents considered synch to be less of a problem in news and football.

Many readers know that European TV is transmitted at 25 frames per second, so a very simple calculation here could perhaps be forgiven. Divide one second by 25 and we learn that each frame of video is on the screen for 40 milliseconds. The accompanying sound cannot have a meaningful delay of less than a single frame, so the 20 milliseconds accuracy demanded by the experts is twice as good as that achieved by today’s broadcast television. One can only praise their desire to raise standards.

Nevertheless, it is a long-established fact in the presentation business that picture quality increases in direct proportion to the volume of the sound. “Play loud” was frequently the written instruction on the audio master in the days when it came in a separate box; it clearly remains true today. The dynamic range (or presumably the lack of it) associated with compressed audio is regarded by those questioned as a sure way to make content “more exciting and involving”, at the expense of true fidelity. Quality is in the ear of the beholder.

The research reveals confusion between the sound and picture quality and the technical issues associated with a poor broadband connection. Unsurprisingly, the interruption resulting from the dreaded “buffering” message appears at the top of the list of unacceptable service issues.

The threshold beyond which the audience starts to dwindle is consistently reported as “two buffering gaps per show”, although viewers of live sporting events are apparently a little more tolerant. After two buffering glitches, most viewers simply stop watching. A great deal of work is being done in this field to improve playback performance, which is clearly needed for paid-for over-the-top (OTT) video to be considered as a reliable revenue generator.

A further misunderstanding is revealed in the answers to a question about the relationship between quality and skipped frames. While it is true that high motion sequences need high frame rates, repeating a frame can be a very effective way of concealing temporary loss of bandwidth. One frame duplicated out of 25 (per second) represents a frame rate loss of 4%, which is almost undetectable except to a trained eye. Yet more than half of those polled demand online performance better than one frame in 100; standards way beyond those common in internet video and only consistently achievable with DVD bit-rates or better.

Surprisingly, in view of the desire to avoid skipped frames, around one third of respondents are happy to throw away five frames in every second by settling for 20 frames per second playback, although this is qualified by suggesting it is acceptable for “animation content and kids’ programmes only.”

With so many technical issues to consider, it is not surprising that certain details were outside the scope of the people running the services. For example, 30% did not feel qualified to talk about buffering and skipped frames and almost 65% threw up their hands in horror when asked to define the “optimum chroma subsampling ratio.”

Cryan concludes that it is not the case that these details don’t matter, simply that “they don’t know because they don’t need to know; their concern for the moment is with the active issues that they need to get right today.”

Quality expectations that viewers bring to their online viewing, however, are about to become much more important as online video moves away from the 20-inch computer screen and on to the 42-inch (and larger) TV. “In general, they will expect video of a much higher standard than we are used to seeing over the web,” says Cryan, 

“If we are going to take online video into the living room, the importance of quality is only going to matter more. That shift in context is likely to be key.”

“What constitutes quality” is a question that has yet to be answered by the ultimate arbiter, the consumer. For the moment it appears to be summed up as, “Quality is whatever I am prepared to pay for.”