Introduction
Asking which format(s) were most popular on the Net a few years ago is a question that will get you different answers depending on who you ask, and where the experts were looking. Arguably, Apple QuickTime format movies were the most popular format during the first few years of the Web's explosive growth. Apple users had an early head start in the digital video field, thanks to Apple's pioneering work with QuickTime and the fact that the first Web browsers and Web authoring tools ran best on Macintoshes, and it took Microsoft until 1996 to release its first version of what was then known as ActiveMovie, accessed via Media Player for Windows 95.
ActiveMovie, in its initial release, supported a number of different types of media, including MPEG (Moving Picture Experts Group) video and audio, Video for Windows, QuickTime (basic features only), and more. For most web users, its MPEG-1 features were its most valuable asset, as there were at that time relatively few other freely available, full-featured, high-performance MPEG players.
By 1996, Real Networks was already taking the lead in an emerging category known as "streaming video," which we'll discuss in more detail later in this article. But, back in 1995 and '96, Microsoft seemed to be primarily focused on its Audio Video Interleave (AVI) format, and quickly advanced its multimedia tools and technologies to keep stride with Apple. With the inclusion of Mediamatics' MPEG player code in ActiveMovie and the huge installed based of Windows users -- almost all of which dutifully upgraded to Windows 95 during that period -- MPEG and Microsoft's own AVI formats grew quickly in popularity. AVI was, for most PC-based digital video enthusiasts, the preferred format primarily because it was more easily edited than MPEG movie files. (MPEG uses a so-called intraframe compression method that makes it difficult to isolate and edit a specific frame.)
Because MPEG-1 and its higher-resolution successor, MPEG-2 are very compact file formats, and players are today freely available for virtually all platforms, MPEG is a good choice for Internet distribution of video and/or audio clips. Net video remains a challenge for dial-up users, though. At 1 or 2K per second, it takes a long time to view even a few seconds worth of MPEG video, and full-screen, full motion is out of the question for all but the lucky few with plenty of bandwidth to spare. A minute of full-motion MPEG-2 video requires 10 megabytes or so of data. (The two most common formats, SVCD and DVD, can store about 45 minutes and 15 minutes of video on a 700 MB disc, respectively.) This translates to a per-second quota that is not out of the question for Cable or ADSL users, but remains out of reach for those with less than 300Kps connections. In short, reasonable quality, full-motion video requires about eight to ten times the throughput of today's fastest modems.
MPEG-1 remains popular in areas other than North America, where the Video CD has gained acceptance as an alternative to VHS video recordings. MPEG-1 videos can be burned onto Video CDs using tools such as Nero from www.Ahead.de (etc.). Interestingly, more than 740MB of VideoCD-format video will fit on a 700MB blank CD.
By 2002, the cost of hardware-based MPEG-2 video encoders from Cirrus Logic and other vendors (notably, ADS) had dropped to a few hundred dollars. The 6-to-8MB/sec data rate of these hardware encoders is below the 12Mbps rate provided by USB1.1, making it feasible to produce low-cost USB-based video encoders such as ADS's USB Instant DVD systems for Windows and Mac.
QuickTime
QuickTime, meanwhile, continued to enjoy popularity among higher-end users, many of whom found that QuickTime movies tended to enjoy better synchronization between audio and video than AVI files of the era. But if your idea of Internet video is a pixilated, herky-jerky video the size of a postage stamp, you haven't seen some of the latest technologies. Although digital video enthusiasts back in the early days of the Web had to put up with blocky 160x120-pixel videos that strained their computer's resources, the latest generation of technologies can truly deliver digital video on PCs that is every bit as good as the best standalone DVD or laserdisc player -- in other words, quality that is superior to that achievable with a VCR or the execrable NTSC North American Television Standard.
Microsoft Windows Media
The big breakthrough for Web video enthusiasts came in 1999, with the release of Microsoft's not-quite-MPEG-4-compliant Windows Media Video compressor/decompressor ("codec"), which the company now supports exclusively via its Advanced Streaming Format (ASF, also known as WMV). A well-produced Windows Media clip can produce, on today's more powerful PCs, video quality that approaches the smoothness and fidelity of a digital video disc. Even more interesting is the fact that the format supports Digital Rights Management (DRM) and a number of other features such as support for interactivity, allowing the user to manipulate individual elements on the screen.
Microsoft in Jan. 2003 set pricing from licensing of its Windows Media 9 technology to other vendors -- at about half the price of competing license from the MPEG-4 Licensing Authority, known as the MPEG-LA. Predictably, the MPEG-4 group accused Microsoft of unfair competition. How dare they undercut the established pricing model?
It is possible to convert some streaming ASF files (currently, not including those encoded with Windows Media Encoder 9) to standalone AVIs, using the freely available "ASFtools." Conversions may occur with or without re-encoding. The author of ASFtools recommends Nimo's Codec Pack as a good "all in one" collection of all the codecs you may need.
Meanwhile, it was undoubtedly to Microsoft's chagrin that the underground hacker community "borrowed" its MS MPEG-4 Version 3 codec and reverse-engineered it (and adding support for MP3 audio), to be able to support formats other than ASF. This led to an underground market in so-called "DivX ;-)" videos -- super high quality wide-screen AVI movies compressed with a codec closely related to the MS MPEG-4 V3 format, that can fit a full two-hour movie onto one or two recordable CDs. This is a remarkable reduction from the disk space requirements of uncompressed video. For comparative purposes, a single stream of professional-quality D-1 uncompressed video requires 22MB per second (!). So-called "Dual-stream" editing systems (which professional editors require) would therefore require 2x 22MB/sec.
DivX eventually dropped its dependence on Microsoft's copyrighted code, and version 5.0x of the DivX code is gaining acceptance as one of the most efficient "mainstream" video codecs, in freely available, "adware" and commercial offerings from www.divx.com. As well, a fully open-source variant, known as XviD also appeared in 2002. Both are optimized for high-quality, compact encoding of standalone or downloadable movies, not streaming video.
Compression and codecs
Even compressed video, using a typical video capture board that saves files as standard AVI or MPEG-1 files, takes a lot of disk or Web Server space. For example, the to save a 1.5 minute MPEG-1 video at reasonable quality with the All in Wonder 128, a popular Canadian-made graphics card with video capture capabilities, requires something like 28MB of storage. DV video, as is captured from a digital video camera over a FireWire connection to a PC or Mac, requires about 10MB per second. You'll need a gigabyte of storage for each 4.5 minutes of DV video.
This means that to create full-resolution, full-quality tapes, you have to capture the video at full resolution, and, if you want decent quality, you must limit the amount of compression used -- or use a vast amount of disk space. Most consumer-oriented analog video capture cards require between 4 and 8 GB of storage per hour of video, depending on how much compression you are willing to use.
And what happens when video is compressed? Compressing the digital video data degrades the image, as the compression algorithms simplify the images by reducing colours, eliminating fine details and reducing the overall fidelity of both the video and audio signals. In some cases, compression may also reduce the number of frames per second from the standard 29.97 fps speed of the NTSC standard. In extreme cases, such as you might encounter when video footage has been highly compressed for low-bandwidth distribution via the Internet, these compression artifacts can lead to blocky, visually unpleasant video clips with poor frame-rates and tinny audio. These "modem friendly" digital video clips have, we think, given digital video a bad name with consumers, many of whom probably assume that all computer video looks like this.
What they don't know is that it is possible these days, for as little as a few hundred dollars, to produce digital video of what might be fairly termed "near broadcast quality." Near Broadcast Quality is approximately what S-VHS or Hi-8 delivers: a better than composite-quality signal, with strong chroma and luma definition. VHS quality is, it seems, a more suspect term that is somewhat of a euphemism for "fairly poor quality." High quality S-VHS or Hi-8 seem to be the terms used to describe the quality you can achieve with 5:1 or better compression at data transfer rates in the 2-3K/sec range. Although you cannot get "true broadcast quality" for $1500 (or even triple that), under-$500 cards such as the above-mentioned ATI All in Wonder 128 or the Matrox Marvel G400-TV should be able to store 20 minutes of near broadcast quality video in less than 500 MB of disk space.
Unfortunately, even highly optimized digital video movies tend to be large. A quarter-screen, "VHS quality" 30-second movie clip designed for Web download will typically be several megabytes in size and can take minutes to download even over a fast connection. Thus, even a highly optimized one-minute movie trailer in QuickTime, AVI or MPEG-1 format represents a downloadable file large enough to deter those with slower internet connections from watching -- or will, at the very least, make them wait a while. Even audio files can be huge. For example, a 7-minute, monophonic sample recorded at 22KHz (half the quality and, of course, half the number of sound channels of a typical stereo CD) results in a file about 17 megabytes in size. Compressing that file as a RealAudio file resulted in a 404K file. Add in some quarter-screen-size, heavily compressed video and you've already all but eliminated real-time "streaming" access by those without broadband Cable or ADSL connections.
It's the dilemma of every Webmaster: he or she doesn't want to keep the audience waiting or lose too many people with outrageously high minimum system requirements. The solution of many is want to take short clips of the movie and re-compress them at a smaller picture size (hence, the term "postage stamp video") and one or more lower data rates. Compression utilities from Real, Microsoft, Sonic Foundry and others allow content creators to specify exactly how much bandwidth they would like their video and /or audio to take up. Most authors produce three versions, for users with 56K (modem), 100K (ISDN) and 250K (Cable or ADSL) connections. More bandwidth, as you might guess, means it is possible to deliver higher frames rates, higher resolutions and/or less compression, yielding better quality images and sound.
Clearly, your visitors have to be very curious about your video content to invest the time, especially when they pay connect-time or per-megabyte charges for their internet access.
There are a number of products currently available that can play a digital video or audio file while it is still being downloaded. This is, in its simplest form, the essence if how streaming works.
Streaming
Some companies, such as the now defunct www.netmoviemania.com, gambled that users would be willing to pay for streaming content. This site, and others like it, built up a substantial library of free and pay-per-view videos from legally licensed materials. Until its demise in the wake of the dot-com crash, users could watch full-screen commercial films, many of which were originally "B" movies and/or several decades old, streamed over the Net with generally "fair-to-good" (sub-VHS quality) results. Many venues now focus on more specialized markets. AtomFilms.com, for example, specializes in short subjects. Elsewhere, you'll find movie sites featuring independent film makers, foreign films, and of course porn -- perhaps the biggest online movie market of all.
The challenge, therefore, is how to deliver "good enough" video to enough people to build a business. One way that this can be done is via a special type of streaming called Multicasting. The main problem with multicasting is that, because the same set of bits is being sent to many recipients, it eliminates much of the freedom and interactivity that makes Unicasts so flexible. With a streaming Unicast, you can pause, rewind, or fast-forward the content you are watching at will -- and it is precisely this freedom that makes Unicast such a bandwidth hog. Every viewer needs to be sent his or her own set of bits. Multicasting sends all viewers the same set of bits, but with this economy of scale, we lose the power for "on-demand" control. See our Unicast versus Multicast report for more details.
In either case, providing video and/or audio without the wait is what "streaming" is all about. As mentioned earlier, Real Networks is currently the market leader in this category, and both Apple and Microsoft are major players. Indeed, a Media Metrix report released in Jan. 2001 claims that streaming media players are now installed on 99% of U.S. home PCs. However, Apple's QuickTime lags far behind its rivals in both market share and capabilities. Insanely Great Mac notes that, as of November of 2000, QuickTime 4.0 was only used by 4% of U.S. home computer users. In comparison, RealPlayer and Windows Media Player 6&7 captured 28% and 22% of U.S. home computer users, respectively. Still, Apple appears to be making some progress against its rivals. Apple announced in Feb. 2002 that there were 80 million downloads of QuickTime Player in 2001 as opposed to 75 million of RealPlayer and RealOne.
Real points to statistics that tell a very different story. According to figures from Nielson//NetRatings, RealNetworks had 32 million home users in the US in December 2001, compared to just 7.4 million users of QuickTime. A previous report from Jupiter Media Matrix saw QuickTime capture 7.3 million users, or 8.9% of Internet users, in January 2001. And that figure was down from 7.9 million in the year-earlier period. Meanwhile, 25.9 million people used RealPlayer (up from 17.5 million in the same month the previous year) and 21.5 million used Windows Media Player (up from 16.4 million).
According to this 2001 Nielson Netratings report on streaming media, Real Networks was, at the time, the leader in the market, with about 21.8 million users per month (the company claims more than 115 million unique registered RealPlayer users worldwide).
By 2002, the figures told a different story in the QuickTime vs. WMP vs. RealPlayer battle. This time, Nielson reported that Windows Media Player had surged ahead with 32 million home users, compared to RealPlayer's 14 million and Apple's 7.7 million.
However, as Mark Twain said, "There are lies, damn lies and statistics." Thus, the answer you get depends on the question you ask -- and who you ask. (Read more....)
The seemingly innocuous question: "What is the streaming video system with the largest audience?" is an example of what might be considered a trick question. In this case, Real's 115 million registered users is a pittance compared to the 150 million users that have browsers capable of supporting Java. So, technically, a Java-based streaming video client provides the largest audience and the fewest compatibility hassles (i.e., no plugins are required).
And, if you look at "enterprise customers," Microsoft comes out on top -- By 2000, it already held a 46 percent market share in this category. And if one were to poll users of computers in translucent plastic cases, it's a safe bet Apple would do well.
However, having a player installed is only half the battle: it may or may not have the codec (compressor/decompressor) required to decode the video and/or audio. Many Windows users have experienced a movie that appears blank, with audio only -- a telltale sign of an incorrect version of the Windows Media codec. Or, the supplier of the video may have used a codec not supplied by default with mainstream players, such as the new AVC codec, known officially as Recommendation H.264 or by its ISO name: ISO/IEC 14496 10 Advanced Video Coding. The technical design of this codec was finalized in Dec. 2002; it will, say its developers, stream video over the Internet at DVD-worthy bitrates. The new compression technology, expected to be ready for public consumption in March 2003, has demonstrated data throughput of just under 1mbps in trials.
Until this codec proves itself, the best bets for streaming Internet video are Microsoft's Windows Media Series 9 or MPEG-4, shipped with QuickTime and supported by a number of video players on various platforms, including the open-source mplayer for Linux and Mac OS X.
The term MPEG stands for "moving picture experts group." This group established a number of standards for audio and video compression, especially as they relate to the development of software or hardware compression algorithms optimized for “digital movies.” In brief, MPEG-1 is approximately a quarter-screen format (352x240 pixels).MPEG-2 is the format used on DVD discs, and is the format transmitted by today's most advanced satellite broadcast systems. Although Apple promised MPEG-2 support would be shipped with the final release of QuickTime 6, it was not present in the version delivered July 15, but costs US$19.95 extra (separate versions are required for OS 9.x and OS X), as a downloadable component, available from Apple's online store.
MP3, contrary to popular belief, does not stand for MPEG-3, nor is it a "free" open standard. It is an audio compression format officially called MPEG Layer 3.) and is based on technology owned and patented by a Germany's Fraunhofer Institute.
MPEG-4, as implemented by Apple in its QuickTime 6 product, is a high-quality format well-suited for streaming video delivery. Although the original MPEG-4 specification was based in part on the QuickTime file format, Apple did not invent MPEG-4; the company merely capitalized upon it early. In fact, a licensing dispute with the MPEG-4 licensing authority, known as MPEG LA, held up the release of QuickTime 6 for several months; the software was finally released in July 2002.
In all its variations, MPEG compression exploits practically every trick in the book to squeeze the enormous amount of data required by full-motion video down to play back at reasonable frame rates and picture sizes on today’s PCs. MPEG-2, as a look at any DVD disc will tell you, improves video and audio quality even further and can be combined with various programming technologies to produce all-digital movies with interactive features and enhanced multimedia features.
The MPEG family of codecs, of course, aren't the best choice in compressor/decompressor technologies for all purposes. In fact, Apple’s QuickTime and Microsoft’s Video for Windows technologies include several “codecs,” each optimized for a particular purpose. Those intended for animation, for example, trade image quality for speed, while still-image compressors favor image fidelity over speed.
The Cinepack codec from SuperMac (licensed to a variety of companies over the years, including Microsoft, Apple, Atari, and 3D0) is one of the most widely used full-motion compressor/decompressors. It was superseded on the Mac by a codec from (and called) Sorenson that delivers the best quality on that platform in QuickTime versions 4 and 5. And the MPEG-4 support in QuickTime 6 delivers even better quality, particularly in the QuickTime 6.3 implementation, released in June, 2003. QuickTime 6.3 also addresses audio/video sync problems -- a common complaint with earlier releases. See our QuickTime 6.3 report for further details.
The buzz about the emerging MPEG-4 video standard had been growing since the Dec. 2001 commitment of support from Real, the founding of of the Internet Streaming Media Alliance (a group that streaming video market leaders RealNetworks and Microsoft have so far declined to join) and the emergence of several free, open-source and commercial DivX variants.
Although Apple was a founding member of the ISMA, it didn't publicly show support for MPEG-4 until Feb. 2002 when, at the company's QuickTime Live conference in Beverly Hills, on Feb. 12, Apple showed a version of QuickTime with support for MPEG-4 video.
However, the company said plans to release this version had been shelved "indefinitely," due to MPEG-4 licensing requirements it said were unreasonable. The licensing terms to which Apple objected were proposed by MPEG LA, a licensing body representing 18 patent holders with claims on underlying MPEG-4 technologies. Under the terms of its proposed royalty model, adopters of MPEG-4 technology would have been liable for licensing fees as high as a millions of dollar per product. But this fee wasn't the deal breaker. Apple said "We are willing to pay, if the technology is worth it." However, MPEG LA also demanded providers of QuickTime streams to pay $0.02 (US) per hour, per stream. It was this fee, not the codec royalties that Apple considered unacceptable. Apple's decision to shelve QuickTime 6 also affected QuickTime Broadcaster, a new program designed to stream live events. See ZDNet.com or CNET News.com for details. These licensing wrinkles kept QuickTime 6 from official release for more than five more months.
In July 2002, the MPEG LA licensing group agreed on what it called "reasonable licensing terms" that, in part, tie royalties for use only to MPEG-4 Visual products or services for which the content or service provider receives remuneration. The licenses require no royalty payments for the first 50,000 decoders and first 50,000 encoders per year targeting Internet (wired and wireless) or mobile users.
However, not everyone considers the terms reasonable. In June 2003, Japan's mobile video content providers threatened to snub the MPEG-4 compression format--touted as crucial technology for delivering video to mobile handsets--unless the cost of using it comes down.
And, although Microsoft is distancing itself from MPEG-4 by asserting that its Windows Media technology is superior, its MP4-V3 compressor/decompressor ("codec") is actually a close cousin to MPEG-4 technology -- and is the codec upon which the original DivX implementation was based. Microsoft in July 2002, began talking up is Windows Media 9 technology, code-named "Corona," in a move many saw as trying to steal Apple's thunder. The company subsequently released the final version of WM9 in Jan. 2003.
Further complicating the issue, On2 has offered its open-source VP3.2 open-source codec to the Internet Streaming Media Alliance as an alternative to the MPEG-4 "standard." As we've noted in the past, this is an excellent codec, unfortunately, hampered by the cash-crunched company's inability to more effectively market it. Indeed, some feel the ISMA should dump MPEG-4, with its onerous royalties, in favour of a truly open standard.
In fact, it can be argued that, on the PC, the open source relatives to MPEG-4 already rule, thanks to powerful implementations from several developers working with free, open-source and commercial DivX variants.
The Indeo 5.0x codec from Ligos Technology (originally developed by Intel Architecture Labs) is still a popular choice for playback on less powerful computers. As might be expected from a compression technology developed by chipmaker Intel, Indeo works best on computers based on the Pentium family of processors. Thus, it is a poor choice on the Mac side. It is, however, capable of serverless video streaming over the Internet -- an angle that Ligos President and Chief Executive Officer Peter Forman is clearly interested in capitalizing upon.
Indeo video's Progressive Download, Ligos says, enhances the user experience because it enables them to begin watching a video download immediately, even at dial-up speeds. While the user is watching, the download continues in the background, with the image quality and frame rate improving as the download and playback continue. When the complete video has been downloaded, quality is uniformly excellent. If they wish, users can then view the clip again at top quality from the beginning. Users with high-speed connections get full-quality video almost immediately. With Indeo video, developers only need to create one video file, which scales automatically to the user's connection rate.
Naturally, due to the inescapable bandwidth demands of high-quality video, you aren't going to get full-screen, full-motion "VHS quality" or better video with sound in much less than about 300K per second's worth of bandwidth, but the latest RealPlayer codecs (which finally achieved what we'd classify as decent performance with the v8 release), Microsoft's MPEG-4-based Advanced Streaming Format, and Apple's QuickTime -- version 6.x of which finally adds long-promised support for ISO-standard MPEG-4 -- are all pushing the edge of what is possible, within the confines of today's (and tomorrow's) Internet infrastructure.
Until the advent of today's fast computers, DVD decoding required special chips with built-in routines for performing the run-length encoding, delta compression, discrete cosine transformations, and other functions that comprise the MPEG-2 format. “Doing it in software” necessarily increases the load on the CPU and, as such, reduces the performance of full-motion video playback, on slower computers. Typically, software-only video playback on an older Pentium 100 allowed, at best, a quarter-screen (352-by-240 pixel) MPEG-1 video to run at 15 to 30 frames-per-second. These machines were, without additional hardware, incapable of playing MPEG-2 files at anything more than a few frames per second. Smooth software-only DVD playback requires a Pentium II 333 or better for smooth performance.
Can you imagine the computer power that is required to display the big-screen version of Don Bluth's Titan AE, touted as the first movie to be delivered to theatres via the Internet in an all-digital format? Reportedly, the digital video file for this movie is a whopping 64 gigabytes in size. Indeed, a whole new generation of high-capacity data delivery systems and projection devices will be required at theatres where George Lucas' plans to deliver the next episode of Star Wars in an all-digital big-screen movie format. This was practically unthinkable a few years ago, but will probably be commonplace in the next few years.
Many people still have 28.8 or 33.6 kilobits-per-second (kbps) modem links to the Internet. A 28.8 modem can, if we do the math (eight bits per byte), download a movie at about 2.5 kBytes per second maximum. Thus, a 10 second, 500k movie can take several minutes to download.
Even if you are creating content for an intranet-only T1 or T3 connection you will find yourself limited by bandwidth. A T1 line can transfer about 100K per second so under the best conditions, a 500k movie may appear almost instantly - but remember that if that a site is running on a 1.4Mbps T1 connection with 10 people all viewing or downloading unique video content at the same time, you are only going to see one-tenth of that bandwidth. Thus, on popular sites, achieving reliable video streaming is a challenge. It is this issue that led Apple to invest in Akamai, a company that specializes in "localized delivery" of high-bandwidth content. Simply put, Akamai sets up servers all over the country and, when you request a video from Apple's QuickTime website (quicktime.apple.com), it comes from the server nearest you, in hopes of minimizing network latency. On the Windows side, a company called NetApp.com provides a similar service, delivering streaming media from the edges of the network to improve streaming media quality.
It's worth outlining how streaming works. In short, the hypertext transport protocol (HTTP) that is used to deliver Web pages to your browser is ill-suited for streaming . This is because, technically speaking, HTTP is known as a "stateless" protocol. The server doesn't know whether the client is ready to receive the data or not. This makes it less than ideally suited for delivering video which, by its nature, depends on feeding the video data to the client at a known rate. There are workarounds, of course. If the video file requires less bandwidth than that which is available, HTTP can open the file and begin to play before it has finished downloading. This leads to at least one notable drawback: the file, which has essentially been downloaded, is then left sitting on the client's hard drive, perhaps in the TEMP directory.
A true streaming server will not do this, potentially protecting the content from unauthorized reproduction. Streaming servers from Microsoft and Real use a "redirector" file that is merely a text file that points to some streaming media content on the server. However, it specifies a different protocol to access this content than the usual hypertext transport protocol. For example, Microsoft's system uses ASX files to point to the ASF files on the server. These ASX files say mms://servername/filename.asf instead of http://servername/filename.asf. Both protocols would load the ASF file, but the MMS protocol requestor causes the Microsoft streaming service to kick in and it delivers the content to the client using Microsoft's Windows Media Technologies and the Advanced Streaming Format. Thus, systems that promise progressive download "streaming" using standard HTTP protocols are never as capable as servers supporting dedicated streaming protocols.
Another question open to debate is which plug-in/player (ShockWave, QuickTime, Windows Media Player, Real, etc.) holds the most promise. It can be argued that Flash, with its bandwidth-friendly "ShockWave Flash" vector format and extensive scripting support is the leading contender in this regard, despite the fact that it is not a streaming video system, per se. It is currently enjoying something of a heyday, as designers use it to produce sites with greater multimedia pizzazz than can be accomplished through standard HTML coding. The fact that Macromedia announced at the Flashforward 2000 conference held in San Francisco, Mar. 27-29/00, that it would offer free software development kits (SDK) to aid developers with integrating Flash into their applications, didn't hurt, either. The Flash Player source code SDK, which is only available to registered developers, adds the ability to port the Flash Player to other platforms while the Flash file format SDK allows for Flash 4 files to be exported. The announcements made during the conference's 90 minute keynote can be viewed at Apple's QuickTime site: www.apple.com/quicktime/qtv/flashforward2000/
Apple, which is, by most accounts, the number three player in the streaming video market, has been hobbled by the fact that both Real and Microsoft's codecs delivered better quality streaming video than Apple's best during most if not all of 2000. To address this issue, Apple on Jan. 24, 2001 posted QuickTime Streaming Server 3 Preview, which added improved streaming capabilities, Web-based administration tools and support for Mac OS X. The company subsequently released QuickTime Streaming Server 4 and previewed QuickTime 6, with improvements in its codecs and MPEG-4 support for Altivec-enabled G4 processors Macs, in Feb. 2002.
But market stats don't tell the whole story. Network Computing did extensive testing of all three platforms in Feb. 2001 and rated RealNetworks' server solutions the easiest to set up and the best overall, but noted that Microsoft's solution was the most full-featured. It rated RealServer an "A," Windows Media a "B" and QuickTime a mere "C-". A subsequent test in March 2002 clearly showed major improvement in the latest QuickTime and Darwin Streaming Server releases.
Says NWC, "After we spent weeks in arduous testing and watching our favorite movie over and over, our numbers showed that Apple's Darwin Streaming Server 4 and QuickTime Player 5 package deserves top honors. Apple proved it could deliver a robust server and client that performed well. Its images beat the competition over a range of bandwidths. Best of all, the software is free, regardless of which operating system you're running. RealNetworks' RealSystem solution took a close second and fell well behind Apple in the price category, since the costs of RealNetworks server software escalate into the thousands of dollars, based on the number of viewers you wish to serve. Finally, Microsoft's Windows Media Services fared poorly in all but the highest-bandwidth quality tests. If you're running Windows 2000 Server and all your clients are Windows-based, this software is a no-brainer, but it's truly the least common denominator in streaming video." See http://www.networkcomputing.com/1306/1306f1.html for details.
It's recommended reading.
Unfortunately, this improvement may have come too late. Insanely-great.com, a Mac enthusiast site, notes that major content organizations are dropping QT support left and right, in favour of more popular technologies from Real and Microsoft. Even ResExcellence.com, another Mac fan site, admits: "to be honest, after experiencing the same content in RealVideo, QuickTime was inferior." But there are wrinkles here, too. As of more than 16 months after the release of Apple's Mac OS X, there was still no Real player for the operating system (the company finally released a beta version of RealOne Player for OS X in July 2002), and accusations that Real has included spyware in its products have affected the player's popularity on Windows, as well.
All of the leading streaming video companies are hard at work to extend their products to provide digital rights management (that's a euphemism for copy protection) features, interactivity, e-commerce hooks and, of course, better video quality.
Microsoft's ActiveMovie Streaming Format (ASF), for example, now allows multiple data objects (for example, audio objects, video objects, still images, events, URLs, HTML pages, and programs) to be combined and stored into a single synchronized multimedia stream. It provides an encapsulation that allows existing popular media types and formats such as MPEG, AVI, .WAV and Apple QuickTime to be synchronized and stored efficiently on a variety of servers. It supports digital rights management, pay per view and supports the emerging "SMIL" interactive multimedia command set that allows web authors to make clickable movies. As, as noted earlier, its licensing model is significantly more affordable to third parties than that of the MPEG LA.
Microsoft's ASF and MPEG-4 technologies, which as mentioned earlier, are becoming to video something akin to what MP3 is to audio, thanks to the underground community, probably carry the greatest potential for market share growth in the short term. We wouldn't be surprised to see a phenomenon similar to that which happened in the browser market where the entrenched leader (in this case, Real) sees its dominant market share eroded by a competitor that not only has greater resources to throw at its products, but currently enjoys the benefit of having its product built into the operating system of the vast majority of the world's computers. Indeed, Microsoft, at the Streaming Media East show in Dec. 2001, previewed its next version of Windows Media Technologies, code-named "Corona," issued in final form in Jan. 7, 2003. At its unveiling, the following key Corona capabilities were previewed:
Whether the software, now slated for a public preview release in Sept. 2002, lives up to its claims or not, it's safe to say that Microsoft will continue to push ahead with its Media Player technology. As Windows 2000, with its built-in support for ASF streaming services, gains hold in the marketplace, as we fully expect it will, Real will likely see its position challenged at the top of the heap.
Real Networks
There are, of course, many other issues -- and not all the challenges are technological. Apple and Real announced a deal in June, 2000 in which Real will incorporate Apple's QuickTime streaming support into its servers (a deal apparently based on the "enemy of my enemy is my friend" principle). Meanwhile, Real also danced with the devil by licensing Microsoft's audio technology for RealJukebox -- a deal that took a twist in July 2002, when the company announced its intention to release its own Windows Media-compatible Helix Universal Server as open source, which led Microsoft to question whether RealNetworks was violating its license terms. According to the New York Times, the company said it "would need to determine whether RealNetworks licensed the software before taking action."
Microsoft had originally portrayed the deal as an acknowledgement that its Windows Media format is an important standard. Real, however, probably sees the long-term benefit in being able to deliver a player to its customers that can replicate the ASF functionality of Microsoft's own. In a July 22 announcement, Real Networks founder and CEO Rob Glaser was careful to avoid using the term "reverse engineering," insisting instead that the company had achieved compatibility with Microsoft's patented codec through strict clean-room engineering, by copying Microsoft's transport and not the codec itself. RealNetworks has already released the client-side code, and revealed some of its server plans. Among the revelations announced in Dec. 2002: Mac OS will be dropped as a server-side development platform.
And in our tests of a Windows Media-compatible stream allegedly delivered by the company's Helix Universal Server, the quality was excellent -- at least as good as anything we've ever seen from Microsoft's own technology and on par with Apple's MPEG-4 streaming in QuickTime 6. Thus, Real, despite complaints from users that its RealPlayer software is bulky, sluggish and surrounded by advertising tie-ins and product placement strategies (you have to watch an ad before virtually every movie trailer on Real's site, and the product's registration process relentlessly pesters you with invitations to third-party offers) is making deals and taking steps that could keep it at the forefront of the streaming field for a while. Through these steps, a future Real Server could theoretically deliver QuickTime, ASF and Real's own media to the next generation of Net-savvy couch potatoes.
Of course, Real isn't the only Open Source streaming solution. Apple's Darwin Streaming Server is open source, and is functionally equivalent to QT Streaming Server. It's also available for FreeBSD, Linux, Solaris and Windows server platforms. However, at this writing, Apple is the only one of the three big players lacking DRM functionality in its software. Real announced digital rights management capabilities for its streaming servers with a technology it calls Helix DRM in Jan. 2003. (Read more....)
There are other license-free alternatives to proprietary formats, too. Ogg Vorbis, a license-free alternative to MP3, made its official 1.0 debut in July 2002, and On2 and Xiph.Org Foundation, developer of the Vorbis format, announced a deal in June to develop and promote the VP3 format as a full-featured, open-source, royalty-free alternative to expensive solutions such as MPEG-4. In any case, we'll be watching.
(Already, a company called ESynch offers an all-in-one media player called ChoiceCaster that can handle files in Real, Windows Media and QuickTime formats. You'll still need the respective players from Real, Microsoft and Apple on your system to provide the codecs and drivers that ChoiceCaster uses, but at least it provides a unified interface. PC World has details.)
Many analysts foresee the emerging category of Internet-enabled TVs, as the market that will finally help interactive video take off. Or not: the AOLTV set-top box announced June 19, 2000 didn't make much of a splash. Microsoft's UltimateTV never even got off the ground. Maybe it will be like VHS all over again: the adult video market has already proven to be a hotbed (if you'll pardon the phrase) of Internet commerce action; perhaps this will be the market that provides the catalyst for "clickable" movies on the Net.
Graeme Bennett is the editor of Video Buyer's Guide (www.videobuyersguide.com). Having held the positions of managing editor, senior editor and editorial director at Canada's two largest computer magazines, Graeme is currently the editor-in-chief at The Techno Zone. He can be reached at Gra...@BennettArts.com
For Further Reading:
Post new comment