|
Feature
August 2000
Squeeze
plays
The tangled tale of Internet video compression.
Erik Sherman, Contributing Editor
A young company comes up with an Internet-related technology that becomes wildly popular. A well-known supplier of operating systems decides to compete and gives its rival software away. The newer company responds by making its version free too. Involved are standards, many companies in the new industry, and the development of the Internet itself.
Sound familiar? Don't be too sure.
This isn't the usual David-and-Goliath story, and it's certainly nothing the Justice Department has been litigating. Yes, Microsoft is involved, but the question isn't which company will dominate so much as which technology will end up on top. The topic is multimedia streaming. The players: Microsoft, RealNetworks, a number of smaller competitors, service providers, and the ISO (International Standards Organization). At stake, how multimedia will travel to PCs, as well as set-top boxes, cellular phones, Internet appliances, and anything else that might get connected to the Internet. And that's a lot.
|
At stake, how multimedia
will travel to anything that might get connected
to the Internet. |
|
Given the world's preoccupation with electronic content, from television to music CDs, it's not surprising that streaming media is becoming popular. MP3 has quickly marched up the evolutionary scale from curiosity to cause of consternation and court action.
Now video, moved over the Internet, has caught attention for ubiquitous video-on-demand, video conferencing, Web-based training, advertising, and many other uses. One catch: Alongside big potential, video involves big file sizes, putting a premium on the speed of Internet connections. That, in turn, puts a premium on compression to make those files as small as possible.
Video compression is nothing new. The cable industry has used MPEG-1 compression, defined by the Moving Picture Experts Group of the ISO, to push as much programming over coaxial cable as possible (see sidebar, "Pick a number"). Every DVD video disc uses MPEG-2 compression to record movies that would otherwise need more than the few gigabytes available on the discs, and digital set-top boxes also use MPEG-2 to increase the number of channels available to the consumer by squeezing more channels into coaxial cable. But both MPEG-1 and MPEG-2 were designed with an eye to dedicated bandwidth—something the Internet does not provide (at least not yet).
"Because the file sizes are 77 percent bigger [with MPEG-1 and -2], that makes [for] extremely long download time," says Rich Martens, marketing product manager for the mobile information systems group of Sharp Electronics. And so, the ISO created MPEG-4, which is optimized for the Internet's relatively low bandwidth. As well, MPEG-4 provides a multimedia combination of audio, video, graphics, and interactivity (see sidebar, "Nuts and bolts").
"At lower bit rates, [MPEG-4] provides much better compression efficiency than MPEG-2," notes Raj Talluri, CEO of Texas Instruments' digital still camera business unit. "In MPEG-2, there's a lot more overhead because there wasn't the need [for low-bit-rate transmission]."
With all the advantages, MPEG-4 would seem a natural market leader. And yet, it is not—at least, not yet. For example, Sharp introduced a year ago a digital camera, called the Internet ViewCam, that was capable of creating MPEG-4 images. The retail price was $699, and the company sold out the initial run of 10,000 units. Although a new version is now available in Japan, the company has no immediate plans to bring the product into the US.
"When you're on the bleeding edge, it takes time to create a market," Martens says. "Certainly the multimedia aspect of this technology is going to take off."
The establishment
No doubt it will. The questions are when and how? A world view both American and PC-centric makes the answers cloudy for the present. Any compression and streaming approach needs support from application developers, so there is software to play a video or multimedia arriving in a particular format.
The PC streaming-video market is heavily dominated by RealNetworks, which provides client (player) software, server applications, and also hosting services. The company was the first in the space, helped establish the category of product, and, depending on whose estimates you believe, owns 75 to 85 percent of the market. What RealNetworks doesn't have is built-in support for MPEG-4, choosing instead its own compression and streaming approaches. However, a number of companies are building decoding add-ons—called codecs—for MPEG-4 that will work with RealPlayer.
For its part, Microsoft offers Windows Media Player, a streaming-media player that comes bundled with Windows 98 and 2000, and which is available for Windows 95. Microsoft also includes a free streaming server with Windows 2000, with no support for other platforms. RealNetworks provides its server products for a wider array of OSs, including Linux, Sun Solaris, and Windows 2000. However, the RealNetworks server can only pump out 25 concurrent video streams for free; any above and beyond that can cost dearly.
"RealNetworks is expensive," says Blair Fuller, chief marketing officer for Streampipe.com, a company that implements media streaming systems. "A thousand streams is about $60,000, depending on the bells and whistles."
Windows Media Player supports and plays MPEG-4 video streams. But according to some industry experts, the implementation is only partial, employing MPEG-4 compression, but not the MPEG-4 approach to streaming. That means it's an open question how well Windows Media Player would work with video streams that fully comply with MPEG-4. (Microsoft did not reply to requests for an interview for this article.)
Then there are Apple's QuickTime products, which account for many of the video clips available for downloading, but not for streaming. "It has not been a viable platform for broadcasting," Fuller says. "It just never took off. The bottom line is [Apple] didn't come up with the platform that could even do live streaming until, what, six months ago?"
That's a shame, according to Derek Arter, manager of compression and encoding services for Streampipe.com, because QuickTime actually yields the smallest file sizes of any approach. Although RealNetworks has recently collaborated with Apple to stream QuickTime through Real's own RealServer 8 infrastructure to RealPlayers on PCs, such actions are probably too little, too late.
Does it matter?
In short, RealNetworks wants to own all streaming markets, Microsoft wants to own all markets, and Apple wants to own all markets, but has pretty much missed the boat because of its incomplete support for non-Mac platforms. And the ISO's Moving Picture Experts Group would like to own the streaming standard. Desire, alone, isn't enough. As with so many other PC technologies, the victor in the video-streaming melee will depend more on marketing, public acceptance, and perhaps government regulation than on cleverness in coding (see sidebar, "Legal maneuvers").
"My argument is that a standard doesn't matter," says Neal Page, vice-president and general manager of the Osprey Technologies division of ViewCast.com. "Since players are ubiquitous, they don't have to interact with one another. So from a consumer's perspective, it doesn't matter." Page thinks the market model in streaming has shifted from the early-adopter phase, where experimenters abound, to "early pragmatists."
"They need to have content on their site to compete," Page says. "MPEG-4 is not quite ready enough to commercialize. I'm going to guess [it needs] three years."
From the PC angle, that may be true. At the same time, looking at multimedia streaming for the view of other potential uses makes standards look much more appealing. And those other uses may become increasingly important over the next few years. As digital television expands its reach, more cable and satellite companies will want to add services to an ever-increasing number of channels. Many of those services will likely include interactive media, either as a delivery mechanism for information, or as an advertising tool.
"It's not clear that the PC-based [video] players will be the dominant ones," says TI's Talluri.
Broadcast companies and manufacturers of set-top boxes are left with a difficulty. Set-top boxes are designed for years of use without the mechanisms of flash memory or downloading to update their programming or functions. In the same boat, you'll find cellular phones, handheld organizers, and other wireless devices whose owners might want to receive streaming multimedia. The capabilities of many such products get fixed at the time they're built. This presents a problem, because both RealNetworks and Microsoft regularly update their offerings, and changes in server software typically make previous media players obsolete.
Then there is the question of operating systems. Windows Media Player works, as you'd expect, with Windows. But the newest version is compatible only with Windows 98 and 2000, and not even with Windows CE, Microsoft's operating system for set-tops and handhelds. RealPlayer is "OS independent," according to Ben Rotholthz, general manager of systems and products at RealNetworks. The newest release comes in versions for Windows 95 and NT, Macintosh, and Linux. Although the company doesn't have versions for embedded operating systems, Rotholthz says the software could be ported "without too much trouble."
|
MPEG-4 would seem a natural market leader. And yet, it is not
—at least, not yet. |
|
That still leaves the issue of updating a video decoder on a set-top or portable. MPEG-4 can address this issue more easily. As the standard evolves, it still has backward compatibility, so an MPEG-4 video decoder included with a set-top box, for example, will continue to work, even if it doesn't gain advantages from later additions to the technology.
"I hear from various other [manufacturers]—of cell phones, set-top boxes—that they don't want to download anything. In environments where the client side is not easily downloadable or upgradeable...standards that don't change are an advantage," says Al Kovalick, chief technology officer at Pinnacle Systems. Pinnacle, which makes video and Webcasting products for both consumers and professionals, doesn't currently use MPEG-4, but is discussing the standard.
“MPEG-4 is not
quite ready
enough to commercialize.
I’m going to
guess [it needs] three years.”
Neal Page, ViewCast.com |
|
In other parts of the world, however, MPEG-4 is having a bigger impact. "European markets are looking at it more than US markets," Kovalick observes. "European markets love standards—are addicted to them. And there's a bigger and better wireless market there."
Yet the receiving technology is only one part of the equation. Whether the media streams are playing on someone's PC, television, or wireless phone, they will be starting from the Internet. The companies that create the streams for clients are largely "platform agnostic," as Fuller puts it, and most interested in the server tools that let them work effectively. At the end of the day, it may be that convenience, and not compliance, is the deciding factor in winning the codec wars.
|
Nuts and bolts
Two features in particular help make MPEG-4 a good fit for streaming applications, especially when the amount of bandwidth is low: object handling and compression.
Video really is a succession of frames, or images, that create the illusion of movement. A major difference between MPEG-4 and earlier versions of video compression is the degree to which it can separately address the frames and visual objects contained within them. Unlike other technologies that recognize objects in video, MPEG-4 doesn't require objects to have rectangular shapes. The technology can treat virtually anything that can be seen as an object.
Moreover, objects can be elements in the original video or additional elements, such as animation or closed captioning. MPEG-4 gives content producers a degree of control previously impossible, including the ability to move, add, and remove objects as the video is streaming. Producers can also provide the audience with any degree of control desired while viewing a stream, making MPEG-4 capable of supporting interactive video.
Like MPEG-2 and MPEG-1 before it, MPEG-4 also provides methods of compressing digital video. Earlier versions of MPEG provided compression to fit video on DVD discs and to allow more channels of content to fit over coaxial cables. MPEG-4 uses compression to move video and audio over the Internet, especially in the face of the generally low bandwidth available to most consumers. Video compression typically depends on the fact that adjacent frames are often largely identical, except for the movement of objects against a background. MPEG-4's ability to manipulate and track objects makes it possible to show the differences by first indicating where objects move and then how they change.
Motion vectors indicate the direction and distance each object moves, and are tiny in comparison to all the information that would be necessary to completely reproduce the object. However, as things move in video, they also change in angle, orientation, color, and other attributes. This is where discrete cosine transforms, or DCTs, come into play. The DCT is a technique that represents images as mathematical expressions. MPEG-4 uses DCTs to indicate the difference between how an object appeared in one frame and what it looks like in another. Again, the mathematical expression occupies only a fraction of the bandwidth that it would take to describe the entire object again.
In addition to visual objects and compression, MPEG-4 also supports multiple video streams. Someone in an audience could receive not only traditional video, but also graphics, text, and sound. MPEG-4 is truly a multimedia streaming technology. The standard also supports uniquely identifying streaming media as intellectual property, providing a mechanism for controlling the use of content.
With all its potential features, the MPEG-4 standard also offers flexibility. A series of profiles—audio, video, and graphic—designate subsets of MPEG-4 tools and capabilities that decoder developers can choose to implement. This means that content developers can always choose a particular profile supported by installed decoders, eliminating the need to upgrade client software in devices such as set-top boxes and cellular phones, which can be difficult or impossible to update each time the standard is enhanced. Simultaneously, the capabilities of MPEG-4 are not frozen by any given decoder implementation.
|
|
Pick a number
MPEG-1, MPEG-2, and...MPEG-4? Don't let the numbering scheme deceive. MPEG-4 did in fact evolve out of the earlier standards, and no evidence of foul play has been found in connection with the disappearance of MPEG-3.
The Moving Picture Experts Group called its original video-compression standard MPEG-1. The organization then developed MPEG-2 as an expanded version to cover digital set-top boxes and DVD video. The group initially intended to use the MPEG-3 name to denote an extension for high-definition television (HDTV). But it eventually became clear that MPEG-2 offered the necessary tools for that task, and "MPEG-3" was never used.
By the way, don't confuse MPEG-3 with MP3. The popular audio-compression format actually grew out of an MPEG-1 technology called Audio Layer 3.
Now get ready for another twist. After MPEG-4, the next related standard in development is MPEG-7. Instead of controlling compression of video, MPEG-7 defines metadata, or information about multimedia content, including moving images, sound, and graphics. In simple terms, metadata is data that describes the content. Like the user-defined tags of XML (extensible markup language), MPEG-7 allows content developers and users to divulge characteristics of the content—such as author, owner, subject matter, or format—that will be useful in identifying that content when it gets stored in a database. Such metadata facilitates categorizing, searching, and re-use.
But what happened to MPEG-5 and MPEG-6? The group considered calling the standard now known as MPEG-7 either MPEG-5 (the next number in the series) or MPEG-8 (which would be the next number in the geometric progression starting with 1, 2, and 4). But because metadata would be so different from anything done before, the group decided to choose something that wasn't a logical continuation, and so picked MPEG-7. And so whim enters even the temple of technology.
|
|
Legal maneuvers
The streaming software market provides some interesting parallels with the browser market. In each, a small company helped to create the market and grew in size as a result: Netscape for browsers and RealNetwork in streaming. Each company obtained a dominant market position and was then challenged by Microsoft. Each company gave away a client version of its software, with Microsoft following suit. A major difference, at least until now, is that Microsoft grabbed the browser-market lead away from Netscape and ended up in federal court on antitrust charges.
While RealNetworks has maintained its lead, the market has changed—perhaps slowly, but perceptibly. Smaller competitors, such as Liquid Audio, have largely faded from view while the Redmond software machine has grown in strength. Although Microsoft and RealNetworks aren't likely to face each other in court any time soon, the Microsoft antitrust case may well have an impact on the streaming-media market.
"The primarily thing to understand is the limited nature of the [legal] remedy," says David J Laing, an antitrust lawyer who is a partner in the Washington DC office of Baker & McKenzie and a former member of the Antitrust Division of the US Department of Justice. In Judge Thomas Penfield Jackson's remedy, Microsoft would split into two separate companies. One would develop and sell operating systems, and another would produce applications. Under this structure, Windows Media Player would fall squarely into the application company, and wouldn't be a feature the OS group could give away.
Until the breakup occurs (if it does), the interim solution would force Microsoft to provide significantly more information about the Windows OSs to application vendors than previously. But there is no provision for Microsoft to provide "equal time" on the Windows desktop.
"Nobody is going to be able to get their splash screen on Microsoft," Laing says. "They don't have to install an icon for Netscape or RealNetworks in the interim." For the time being, Windows Media Player, and its implementation of MPEG-4, remains unchallenged as the first streaming-media player available to a Windows user.
What would happen after a break-up is anyone's guess. Over one point, however, there's little question. Should the Jackson remedy be upheld in the appeals process, RealNetworks will have a powerful weapon. According to one industry insider who requested anonymity, RealNetworks has already started using Microsoft's own tactic of FUD (fear, uncertainty, and doubt) against Microsoft, hinting that customers should think twice about using a streaming format from a company whose future is uncertain.
The lawsuit also remains a viable tool. "I think Real very likely could [use legal challenges to Microsoft]," Laing says. "The government is going to be looking over [Microsoft's] shoulder, also. The Reals of the world will always have a willing advocate in the Justice Department."
|
Author information
Erik Sherman, a writer and photographer based in Marshfield, MA, ripped the lid of set-top boxes in our July issue. His latest book is Home Networking Visual Jumpstart (Sybex; 2000).
|
