A lasting compression standard?

Given the incredible change in pace of all things digital, one might speculate about the half-life of any technology that can benefit from the ongoing
Author:
Updated:
Original:

Given the incredible change in pace of all things digital, one might speculate about the half-life of any technology that can benefit from the ongoing geometric progression in processing power, memory and storage capacity, network bandwidth, and so on. For example, the following question came up recently in the OpenDTV Forum, an e-mail discussion forum on digital television issues:

“Is there a Moore's Law regarding codec efficiency, or is there a theoretical limit to the improvements we can expect in digital compression algorithms? If so, then how far away from that theoretical limit is MPEG-4/AVC (aka H.264)? Is MPEG-4/AVC to the point that it really could be a standard that could last for 20 years?”

Lasting compression

Video compression has been a fact of life since television hit the airwaves about seven decades ago. One might ask how this is possible, because digital television broadcasting has only been a practical reality for the last decade of those seven. The answer is simple: Video compression need not use digital signal processing techniques to reduce the amount of information that is transmitted.

The true measure of compression efficiency lies in the ability to reduce the amount of information delivered to a receiver in a manner that limits the distortions and artifacts perceived by the viewer. One measure of the theoretical limit faced by any compression technique is known as the rate/distortion boundary (in layman's terms, how much we can squeeze before the distortions become too objectionable to the human observer).

Given the high level of compression artifacts seen today on virtually all digital television distribution platforms, one could properly surmise that this limit is exceeded on a routine basis. Then again, for those of us who have been around for some or all of the past seven decades, we know limits on image quality, artifacts and distortions have always been a problem for television broadcasters.

Over those decades, the analog television standards used throughout the world have relied on analog compression techniques to bring moving images into our homes. Interlace is a powerful compression technique, reducing the amount of information transmitted by half.

Our analog broadcast standards place hard limits on the frequency response of the analog signals that are transmitted. For several decades, it has been possible to capture significantly more detail in 525- and 625-line television systems than can be delivered via NTSC and PAL. This became obvious in the '80s when analog component video processing gear became a practical alternative to composite signal processing. It happened again when analog component outputs from DVD players made it possible to deliver the full quality possible with 525/625-line component video to consumer television displays.

When we added color to TV, we had to squeeze even harder. Where one black and white signal once filled an entire 6MHz to 8MHz channel, we now needed to deliver three components. Only a tiny fraction of the color information from a camera is delivered using NTSC and PAL compression.

While these analog compression techniques imposed hard limits on delivered image quality, we became accustomed to the inherent distortions, even as television images improved over the years. It took about five decades to fully exploit the capabilities inherent in these analog television standards.

In contrast, it took less than a decade to fully exploit MPEG-2, the first digital compression standard used to deliver television pictures to the masses. Now the question is whether MPEG-4/AVC (aka H.264) can be expected to last as the standard for two decades.

The half-life of a standard

Analog television is still going strong. While President George W. Bush just signed legislation setting February 17, 2009, as the official shutoff date for analog television broadcast in the United States, many industry observers question whether this deadline will be any more real than others that have come and gone.

To be certain, devices that will support the signals flowing through that little RCA connector will not disappear for decades to come. Analog cable, analog television translators and huge libraries of analog programming will not disappear on any certain date.

The same is likely to be true for MPEG-2, MPEG-4/AVC and whatever comes next. Legacy standards will live for decades, even as they are replaced by the next generation. A useful threshold to consider when thinking about the half-life of a digital compression standard is that point in time when it becomes possible to realize a 2:1 improvement in compression efficiency, i.e., the ability to deliver the same picture quality with half the bits.

Gary Sullivan, a video architect with Microsoft, has worked extensively in the ITU and MPEG on video compression standards. As chairman of the Joint Video Team, a collaboration of the ITU and ISO/MPEG, Sullivan played an important role in the development of MPEG-4 part 10 (AVC), which is also known as ITU standard H.264.

When asked how far today's technology is from the theoretical limits on compression efficiency, Sullivan replied, “I think nobody knows the answer for sure.” However, he is confident that we can achieve at least another factor of two in practical compression efficiency improvement for video (i.e., a 50 percent reduction in bit rate with similar video quality relative to a good use of H.264/MPEG-4 AVC syntax).

But Sullivan cautions that it may take a number of years before we figure out how. His guess: between five and 15 years. That puts the estimated time necessary to get the next solid factor of two in coding efficiency into the same 10-year ballpark as the last time around — with MPEG-1 and MPEG-2 standardized in 1993 and 1994 and H.264/MPEG-4 AVC standardized in 2003 with its fidelity range extensions in 2004.

Out of sync

So here we are, some three years from the date that analog television broadcasting is supposed to end in the United States, and broadcasters are being told that their new digital television standard, built on MPEG-2 video technology, is already dated. The direct broadcast satellite services are migrating their customer base to H.264/AVC, to take advantage of the bandwidth savings so that they can deliver (H)DTV broadcasts to most local markets across the country. Cable is certain to follow.

H.264/AVC is being deployed rapidly for Internet download applications. It is supported natively by Apple's QuickTime media architecture and is the format used by the company's video download service and iPods. Most videoconferencing systems and some key cellular and mobile TV services are using one of the low complexity profiles of H.264/AVC.

H.264/AVC is not the only new game in town. There is the new SMPTE 421M VC-1 standard based on Microsoft's Windows Media 9 codec design, as well as proprietary codecs such as On2's VP series that are popping up in places like Flash, AOL and Skype.

By the time broadcasters are expected to turn off those NTSC transmitters in 2009, it is likely that they will be a generation behind their competition. To make matters worse, the ATSC standard has poor support for the one aspect of over-the-air television where broadcasters have a competitive advantage over tethered services: the ability to deliver bits to portable and mobile receivers.

Attendees at next month's NAB2006 will have the opportunity to see how much the landscape of digital television has evolved in just one decade. And they will have the opportunity to see where we may be in another decade, as today's emerging technologies will have run their course and begin to yield to whatever comes next.

Broadcasters might question where the opportunity is for them in all of this. Here's a clue: It has nothing to do with government subsidies for set-top boxes that implement dated technologies.

Is the broadcast industry's glass half empty or half full? Have we reached the theoretical limit on squeezing more profits from broadcasting as we know it? What do you think?

Craig Birkmaier is a technology consultant at Pcube Labs, and he hosts and moderates the OpenDTV Forum.

Web links

Description of H.264/MPEG-4 Part 10 video compression algorithm http://en.wikipedia.org/wiki/H.264

Description of SMPTE 421M/VC-1 video compression algorithm http://en.wikipedia.org/wiki/VC-1

MPEG Industry Forum Tutorials on compression technology www.mpegif.org/tutorials.php

Apple Computer H.264 in QuickTime 7 www.apple.com/quicktime/technologies/h264

Microsoft description of VC-1/Windows Media compression algorithms www.microsoft.com/windows/windowsmedia/forpros/events/NAB2005/VC-1.aspx

ON2 video compression technologies www.on2.com

Send questions and comments to:craig.birkmaier@penton.com