It's a Visual World: Putting Video on the Network
Old compression standards never die. They just fade away. Of course, fading away can take the better part of a decade (or even longer), creating a field of technology standards that just keeps getting bigger.
Old compression standarss never die. They just fade away. Of course, fading away can take the better part of a decade (or even longer), creating a field of technology standards that just keeps getting bigger.
Take MPEG-2, which is still widely used in consumer and pro applications, including cable TV and DVDs, even though it's 14 years old. The result is an enormous installed base of devices, many of which won't be replaced anytime soon. So when cable operators, broadcasters, and pro AV vendors want to upgrade to a new standard, they have to respect the past by continuing to support legacy devices and the legacy standards inside them.
CNN Communications Center, New York. Video integration by McCann Systems.
It's a situation that raises a few questions: Are yesterday's compression standards up to today's tasks, such as high definition (HD)? And when picking standards to build into their products, how do AV equipment makers know which ones will have staying power? How do you know which ones to specify for an AV job?
“The most important thing is to understand whether there's a market for it,” says Sean Lessman, senior director of advanced technologies at New York–based Tandberg, which makes videoconferencing systems that rely on effective compression. “We say ‘no' more often than we say ‘yes' because there's a lot of technology out there that people introduce simply because they want to enter a market and can't compete with the giants there now. They want to do something different, shake things up, but it isn't a technology that's going to last. You see that a lot—more often than you see a technology that's going to benefit or revolutionize the industry.”
So where are today's compression standards in terms of their ability to revolutionize the AV industry?
MPEG-2 LIVES ON
Video compression leverages the high likelihood that in a movie or a videoconference, many parts of an image usually change slowly or keep recurring. By processing only the differences between each frame, it's possible to squeeze video down to the point that it requires less bandwidth than the raw, uncompressed content.
Just how much less depends on the compression standard, but the common denominator—not just across standards, but also across the industries that use them—is that they're all a response to the fact that bandwidth rarely is unlimited and free.
“Any industry that would benefit from putting more stuff into something for less money, compression is very important to them,” says Pete Nutley, director of global product marketing at Tandberg.
One of those industries is television—including cable and telecommunications TV services, such as Verizon's FiOS, and satellite services, such as DirecTV. The television industry has spent the past decade flooding the market with MPEG-2 devices, and although the standard has shortcomings, it's nowhere near ready for retirement.
“You're going to see it around for a long, long time because that's what the broadcasters are broadcasting HD in,” says John Lopinto, president and CEO of Hauppauge, N.Y.–based Communications Specialties. “Just like we've had NTSC [the National Television Systems Committee standard] for over 50 years, I think we're going to have MPEG-2 for a long time.”
“MPEG-2 isn't going to go away anytime soon because there's such a huge deployed base,” says Ajay Luthra, U.S. chair of the Moving Picture Experts Group (MPEG) standards body.
Even so, many argue that MPEG-2 has flaws, particularly in terms of the amount of processing required. That's because MPEG-2 just breaks an entire image into blocks, whereas newer, object-based compression/decompression algorithms (codecs) focus on objects—a tree, a house, a face—when deciding where and how to compress. As a result, object-based video coding, which is used in standards such as H.264 (see “H2.64: Horsepower at a Price,” below), is much more bandwidth-efficient than simple block-based coding, even though the objects are eventually broken into blocks.
“If the broadcasters had to do it all over again, I think they'd use [H.264] instead of MPEG-2 because it's much more efficient in its coding algorithm,” Lopinto says. “It ends up creating a lot less bits to transmit the same quality video.”
With AV demands skyrocketing, applications like videowalls, digital signage solutions, or advanced operations centers require efficient compression technology.
However, there are a few applications where compression isn't a must-have, and it's in those areas that the broadcast industry also exerts its influence. One example is the new, third-generation (3G) version of the High Definition Serial Digital Interface (HDSDI), which is a potential fit for live events that don't have to be recorded and thus don't require compression.
“I think you'll see the use of HDSDI more because it's a very robust, single serial stream,” says Lopinto. “If you need to do 1080i, you can do that on an uncompressed basis very simply over a piece of coax for hundreds of meters. You can dual link that to get 2K [resolution] screens. We're still a year or two away from 3G being implemented on a wide basis. I'm not aware of any projectors with 3G inputs on them, but they do have HDSDI inputs, either standard or as an option.”
H.264: HORSEPOWER AT A PRICE
Also known as MPEG-4 Part 10 Advanced Video Coding (AVC), H.264 was co-developed by MPEG and the International Telecommunication Union (ITU) and finalized in 2003. H.264 provides the same video quality as MPEG-2 but at about half the bit rate. That's why it's already built a significant following in industries like as pro AV, where it's available in equipment such as videoconferencing systems.
“All the advanced systems today are MPEG-4–based,” says Jim Smith, CTS, CVE, consulting systems engineer for AV channels at Pleasanton, Calif.–based Polycom. “There is a broadcast industry counter-push to reinforce MPEG-2 as a viable coding scheme, but the data-channel size required is double or triple that of MPEG-4 for the same—arguably—video quality.”
“Videoconferencing is moving very fast toward it,” says MPEG's Luthra. “Adoption of AVC in telco IPTV [Internet Protocol television] is already happening, primarily because they have more bandwidth constraints. It's already happened in satellite. Cable so far hasn't deployed it in any big way, but I keep hearing that they want to do that within a year time frame.”
But H.264/AVC's benefits come at a price: Its sophisticated techniques—such as predictive coding, variable block-size motion compensation, and variable block-size integer discrete cosine transform—require sophisticated hardware. That can increase the cost of equipment, which poses a challenge when selling into price-sensitive markets.
One alternative is the Microsoft-backed VC-1 compression standard—which has the backing of the Society of Motion Picture and Television Engineers, but not the same professional AV profile as AVC (see “VC-1: That Other Compression Standard,” page 54). VC-1 requires less horsepower, but H.264/AVC backers aren't yet worried.
“In AVC, there's no movement—at least formally—to create yet another profile with a simpler hardware,” says Luthra. “People are discussing it, but right now, it's kicking the tires. If they do, that will be a different standard. H.264/AVC is all done and frozen.”
But if 40-plus years of Moore's Law is any guide, hardware capable of meeting H.264/AVC's requirements will become more widely available and less expensive. In fact, that trend is already playing out in consumer AV.
“Today, there are video cameras coming out with AVC encoders,” Luthra says. “These are consumer-grade, HD cameras that fit in your pocket. They have strict power requirements. The complexity is becoming less of an issue because of Moore's Law.”
H.265: COMING SOON?
Even as H.264 builds market share, the industry is already developing its successor, often referred to as H.265. The ITU currently has a work group hashing out the standard, and it will be years before vendors can start building products around it.
Tandberg's new Codian MCU 4500 HD voice and video bridge uses H.264 to deliver 720p at up to 30 frames per second.
“This is slated for publication and review around the end of the decade,” says Polycom's Smith. “With any luck, it will actually contain mandates for the HD interoperations that are left nebulous in H.264, [such as] 16:9-to-4:3 translations, pixel-count translations, and color-depth translations.”
Time (a lot of it) will tell whether it does.
“We don't anticipate coming to market probably for another four or five years,” says Lessman at Tandberg, which is one of the companies developing the standard.
Others anticipate an even longer road to commercialization. “If H.265 happens, it's not likely, in my opinion, to happen in less than five years,” Luthra says. “Commercial deployment will take even longer than that.”
One reason for that outlook is that H.265 would have to be a dramatic improvement over existing standards in order to win support among equipment vendors and other companies. If it isn't, some think that there's a chance that H.265 might never come to market.
“What we learned in the MPEG-2 and MPEG-4 days is that it's not too exciting to do just 15 percent to 30 percent coding efficiency improvement,” Luthra says. “It has to have a 60 percent improvement in coding efficiency.”
MANY DEVICES, MANY CAPABILITIES
Some vendors hope that H.265 can deliver those improvements but without requiring high-end hardware.
“It's got to be able to run on very low-end hardware,” says Tandberg's Lessman. “So there's a lot of effort in that direction.”
The focus on low-end hardware isn't just about cutting costs. In the case of videoconferencing, it's also about accommodating the growing variety of endpoints. For example, as more companies have employees who work from home or spend most of their workday in the field, the ability to extend videoconferences to laptops and even cell phones becomes an asset.
Some cell phones or “smartphones” aimed at the enterprise market are designed to be laptop replacements, depending on the tasks that employees perform in the field. Although some of the latest smartphones have processors in the 500 MHz range, that's a far cry from the horsepower in desktop PCs and other endpoints used in videoconferences.
Even so, mobile videoconferencing services, such as AT&T's Video Share, and mobile videoconferencing products, from vendors such as Tandberg, show that videoconferences will increasingly span a wide variety of disparate devices, each with different capabilities. Those capabilities determine which compression standards they can support.
In those environments, one solution is to use a “light” standard—in terms of processing requirements—such as MPEG-4 Part 2 Simple Profile, which is the basis for applications such as Apple's Quick Time player.
“It's nowhere near as efficient as AVC, but if you don't have the processing power, that's the one you stick with,” says MPEG's Luthra.
But as smartphones become more powerful, their ability to run more sophisticated codecs increases.
“The cell phone integrated circuit (IC) providers are now integrating AVC,” Luthra says. “But right now, if you're doing HD videoconferencing with AVC, and the cell phone can do only MPEG-4 Simple Profile, then you need to translate in between.”
THE SUN SETS FOR SOME …
A few compression standards are starting to fade. For some, it's because they're long in the tooth and can't support today's demanding applications. For others, it's because they never caught on, particularly with equipment vendors. Examples of both include MPEG-1, Vector Quantization (VQ), Fractal Coding, and Wavelet Coding.
“MPEG-1 just doesn't have the horsepower to support modern video resolutions,” says Polycom's Smith. “VQ was designed to run on a x286-based personal computer and doesn't render very well. Fractals showed enormous promise initially, but the coder latencies were so large that it was impractical for everyday use. Wavelet compression was similar to Fractals but featured a slightly different technique.”
Others believe that although MPEG-2 still has years of life, its expansion will begin to slow as AVC builds market share. And MPEG-2 isn't alone in that regard. H.261, a 1990 ITU standard for transmission over ISDN lines, and its successor H.263 are long in the tooth.
“H.261, I have a feeling that its future is in doubt,” Luthra says. “H.263, how long will people use that? They'll migrate to AVC very soon as soon as the CPU power is not an issue.”
Another standard whose days are numbered is MPEG-4 Part 2 Advanced Simple Profile (ASP). “I think Advanced Simple Profile is going to be used less and less,” Luthra says. “It's practically not even used anymore in North America.”
Luthra believes that MPEG-4 Part 2 ASP's sibling, MPEG-4 Part 2 SP, is likely to stick around because it has lower CPU requirements than ASP. But within five to 10 years, even SP likely will be replaced by MPEG-2 or AVC, Luthra says.…
AND RISES FOR OTHERS
One wild card is China, which has spent the past decade showing a preference for homegrown standards—not just for video compression, but for wireless technologies such as cellular and 802.11 Wi-Fi. In all of those industries, the main reason is that Chinese vendors pay lower or no royalties to use a homegrown standard, giving them a competitive edge over products based foreign standards that carry a royalty.
One example is the Audio Video Coding Standard (AVS), China's rival to H.264/MPEG-4. Foreign AV vendors can't ignore AVS mainly because doing so risks sales into one of the world's largest markets. Vendors that don't sell into China also can't ignore AVS because the Chinese market is large enough that its equipment volumes could eventually push AVS into foreign markets.
“I suspect that with time, that [ASP] will go away, especially when China is developing its own AVS standard,” Luthra says. “It's quite possible and likely that they'll say, ‘If you want to do business in China, then you have to support this. Outside of China, so far it hasn't caught on. But time will tell.”
Tim Kridel is a freelance writer and analyst who covers telecom and technology. He's based in Columbia, Mo., and can be reached at firstname.lastname@example.org.