DVD PRO Conference
V2B Conference
TechVideo Expo
Current Issue
Article Archive
News Archive
Buyer's Guide
nbsp; Home Magazine eNewsletters Events Contact Navigation

Current Issue Current Issue
Buyers Guide 2001

Buyers GuideCompany SearchProduct Search

News Indices

CD TrackerCD/DVD-ROM IndexFact, Figures & FindingsConference Calendar

Tech Video Expo

NEW! 2002 Online
Buyer's Guide

DVD TodayDigital EarfulTechVideo NEWSSubscript
Ad Links

Whatís It Good 4: Encoding and Applying MPEG 4

Jeff Sauer

When it comes to low-bandwidth streaming video, todayís common solutions are RealVideo, Windows Media, and QuickTime with Sorenson compression. Served up over a dial-up connection of 28.8 or 56K, they all look pretty lousy. With wireless bandwidth less than half of that, howís MPEG-4 going to make images any better? Has the MPEG committee found a magic bullet for compressing video?

Probably not, but MPEG-4 has some notable advantages over current technologies that could make it the file format of choice for wireless devices and a range of other digital media solutions. Set-top boxes, videophones, digital cinema, and television over the Web are all candidates for MPEG-4 implementation due to its standardization, interoperability, and image quality. And since MPEG-4 is scalable, itís possible that the same video file could be streamed to different devices regardless of individual bandwidths and played back with different degrees of quality. Whatís more, MPEG-4 is not limited to straight video and audio. Efficient still images, 3D and 2D graphics, and interactivity all offer a wealth of product possibilities.

Of course, MPEG-4 is very new, and today very few actual MPEG-4 products exist. Market confusion caused by the non-compliant, shareware format ìDivXî could spell disaster in the battle for public awareness. Yet true public acceptance comes from applications and products that offer a consumer benefit, and several companies are already working to exploit MPEG-4 with potentially powerful solutions that could be on the way as early as this year.


Whatís MPEG-4?

Weíve seen the MPEG acronym before. MPEG-1, MPEG-2, and MPEG-4 are all the intellectual products of the Moving Pictures Experts Group, a subcommittee of the International Standards Organization (ISO). MPEG-1 was adopted almost a decade ago to put VHS- quality digital video onto 1X CD-ROMs and Video CDs, and is now widely used for interchanging desktop video clips, including MPEG-over-IP applications like distance learning, telemedicine, and videoconferencing. With MPEG-2, the MPEG committee focused on better quality video without the constraint of bandwidth. Frame rates were increased to 60 fields per second and typical native resolutions to full screen, and today MPEG-2 is the standard video format for both DVD-Video discs and digital television.

MPEG-4 was conceived for Internet bandwidths below MPEG-1 at between 100Kbps and 1Mbps (bits per second), but in the big picture MPEG-4 can cover a much broader range, going both higher and lower than either of the two preceding MPEG flavors. Yet, as was the case when MPEG-2 followed MPEG-1, itís a mistake to see MPEG-4 as a replacement for either.

To understand MPEG-4, letís start by dividing it into two separate parts. First, the MPEG committeeís primary target for MPEG-4 was to create effective low-bit rate video compression. And just as with MPEG-1 and MPEG-2 before it, MPEG-4 uses the Discrete Cosine Transform as the basis for video compression, also adding the powerful motion predication on interframes that make the other MPEG flavors so efficient. MPEG-4 differs from the others by measuring video over time rather than by a rigid 29.97fps (frames per second), allowing both reduced frame rate video (and thus stretched MPEG GOP or Group of Pictures structures) and even irregular and changing frame rates within the same stream.

Nonetheless, MPEG-4 uses the same basic compression principles as beforeóalbeit fine-tuned and updatedóand therefore does not take a giant unforeseen leap forward in image quality.Ý Interestingly, current implementations of MPEG-4 encoders are arguably no better, if not of lesser quality, than existing technologies like RealVideo and the Windows Media codec at delivering streaming video through tight Internet pipes.

The early attractiveness of MPEG-4 to the industry, including cellular phone makers and content distribution organizations, is that MPEG-4 is a common standard not owned by a single company. That is not the case with RealVideo, Windows Media, or QuickTime/ Sorenson, who leave content distributors potentially owing royalties whenever they are used and, worse, at the mercy of another companyís future decisions. For large streaming video outlets, that could mean serious licensing fees as well as awkward relationships.

Stan Woodward, vice president of Business Services for Yahooís Broadcast. com, says that the company is ìformat-agnostic,î meaning not format-specific. ìWe continue to work closely with RealNetworks and Microsoft on improving their proprietary formats. But sure, weíd love to see an open standard, and weíre also quietly working with MPEG-4 vendors to support their efforts in that direction.î

Ed Hansch, director of marketing for MPEG-4 encoder-maker e-Vue, agrees that the appeal of standardization will drive early adoption of its tools. Their forthcoming encoder will leverage only a fraction of the MPEG-4 standard, with quality Hansch calls only ìcomparable to other streaming codecs.î But Hansch also boasts that e-Vue is receiving ìgreat interest from potential partners, becauseÝ MPEG-4 is an open standard and not owned by any one company.î


The Second Part:

MPEG-4ís Bigger Picture

e-Vueís video encoder is not at all alone in just scratching the surface of the MPEG-4 potential capabilities. Almost all of the early MPEG-4 encoding products announced to date deal expressly with MPEG-4 target of low-bandwidth video and audio compression. However, the MPEG-4 file format can go well beyond what we currently think of as Web-delivered video. Much like the QuickTime file format on which it is based, an MPEG-4 stream can contain several different objects or layers (in QuickTime, Apple sometimes refers to these as tracks), composed of a variety of media and data types. Specifically, in addition to video and audio, an MPEG-4 stream could also include still images, 2D and 3D graphics, non-rectangular video, metadata about the stream (identifying descriptors for more versatile searching), and interactivity, potentially yielding a very rich user experience.

For Rich Mavrogeanes, CTO of VBrick Systems, this part of MPEG-4 will be far more powerful in the long run. ìMPEG-4 has compression, but more importantly, MPEG-4 can do video layers and multidimensional objects, and that makes it extremely powerful for low-bandwidth applications.î VBrick already works with MPEG-1 and MPEG-2 for its network appliance applications and is currently ìexploringî MPEG-4, according to Mavrogeanes, but has yet to make any product announcements.

Optibase is another leading manufacturer of MPEG hardware encoders that has announced MPEG-4 plans. According to Optibase VP of marketing and business development, Yaniv Garty, forthcoming versions of Optibase's streaming video serversódue to ship later this yearówill incorporate MPEG-4. "We see the MPEG-4 standard as having great potential to break through the broadband barrier and fulfill the promise of streaming interactive video-enhanced applications.

In addition to audio/video compression, the MPEG-4 standard also specifies compression types for each of the other object layers. For example, MPEG-4ís native still images use a wavelet-based compression that is three to five times more efficient than JPEG, according to e-Vue, whose currently available ìImage Studioî compresses MPEG-4 stills. And, because Wavelet compression is inherently scalable, low-detail versions of images can load very fast in a browser with increasing sharpness as more data arrives. MPEG-4ís 3D animation is very much like VRML.

By mixing still images, 2D graphics, and 2D and 3D animation with video, MPEG-4 enables characters and objects to be placed in easy-to-compress artificial environments. This capability creates wild new design possibilities and dramatically reduces the amount of data required to transmit video streams. MPEG-4 also frees video streams from the standard rectangular format of television. MPEG-4 videos can be square, oval, octagonal, or even the shape of a face through the use of a one-bit mask around the desired shape.

Imagine a person talking in front of a static background. An MPEG-4 encoder could use that mask around the desired object or shape, thereby eliminating it from display or replacing it with some other layer such as a still image or computer-generated environment.

Ganesh Rajan, IVastís director of Technical Marketing says, ìWeíll do all those features, but one piece at a time. We hope to generate interest with a freeware version of our encoder that just does video and audio, then begin to offer more features as fast as we can.î Rajan believes that once people realize that they can create interactivity with video through hotspots and buttons, or add other non-video media types like charts, graphics, or maps, theyíll be hooked on MPEG-4.


The Big Bit Bucket

MPEG-4 goes even further than its low bitrate target by supporting previous versions of MPEG and, thus, can achieve much higher data rates. While the MPEG-4 standard lays out specific native compression types for its various components, it ultimately has the ability to compartmentalize a wider variety of media objects. For example, the video component of an MPEG-4 stream could contain a very high-bandwidth MPEG-2 file rather than a low bitrate Internet file. Itís this aspect of MPEG-4 that makes other implementations like set-top boxes and digital cinema possible.

Set-top box manufacturers expect to exploit MPEG-4ís data handling ability for two-way communications. An MPEG-4 decoder can send information back to the encoder, therefore video hot-spots or button graphics could yield the DVD-style interactivity that would be required by a set-top which currently might require a separate modem connection. Of course, set-top boxes connected to consumer television sets will need higher bandwidth and resolution than typical MPEG-4 streams, and will use either MPEG-1 or MPEG-2 video within the MPEG-4 structure.


The Wireless Revolution

Today, one of the biggest pushes for MPEG-4 comes from the wireless industry, targeting both cellular phones and PDAs as receivers for video data. Cellular phone makers, wireless carrier companies, and content owners all have an interest in making wireless video happen. It means more data and more billable airtime and itís yet another distribution outlet for existing content, but it also potentially opens to new types of video content and therefore vast new business opportunities.

So-called ìnanny-camsî could allow parents to see their children anytime in the home or at daycare centers. Highway cameras would allow motorists to see traffic ahead before they see the taillights. And doctors could make house calls again, via video.

Unfortunately, todayís generation of cellular has a bandwidth of only 14.4 Kbps, half that of relatively antiquated 28.8 modems. Since more than half that bandwidth is needed for speech, thereís very little room left for video, never mind the 100KB targeted by the committee for MPEG-4.Ý Yet again, MPEG-4ís flexibility has allowed wireless proponents like Packet Video to reduce resolutions and frame rates to achieve demonstrable video results on wireless devices.

Ed Suski, senior director of technical marketing, reports that many employees of Packet Video already carry video-enabled prototype cellular phones or PDAs today. Packet Video, says Suski, also currently ìhas trials going on with 16 wireless carriers in 11 different countries, including Sprint here in the United States and Sonera, the largest wireless provider in Finland.î

Packet Video is leveraging two basic visual profiles of the MPEG-4 standardó the ìsimpleî and the ìsimple scalableî profilesóto deliver video to mobile devices. The simple profile includes a minimal degree of error resilience that importantly ensures the integrity of the video stream. However, itís the simple scalable profile which allows a decoding device, like a cellular phone, to tell a video server what kind of technology and bandwidth it has to work with, thereby allowing the server to send the proper stream for display. Even better, unlike RealVideoís SureStream technology or QuickTimeís ìhinting tracksî, the server does not need to have multiple bitrate versions of a video stream on file, but rather can prune from a single existing stream by reducing frame rate or resolution on-the-fly to accommodate a lower-bandwidth decoder.

And while todayís wireless bandwidths are scant, the next generation of cellular technology, so-called ì2 1/2 Gî (a ìdot revî in computer terminology) is expected to boost wireless bandwidth to 64-144KB this year. A third generation is already in planning that would raise wireless bandwidths to as much as 2Mbpsóplenty for video, but that advance will require wholesale infrastructure changes that are likely to take some time.


I want my MPEG-4

Alas, unless you work at Packet Video or one of the other MPEG-4 pioneers, you may have to wait awhile for a nanny-cam or video cell phone. Like any emerging technology, it will take some time for an infrastructure to get into place and for consumer momentum to grow. However, many industry players believe that products will begin to appear this year.

Whether it happens this year or next, there is already significant momentum behind the scenes from several very large companies supporting MPEG-4 in several areas. Packet Videoís Suski notes that it has had interest from big names like Sony Pictures, Universal Studios, Warner Bros. New Media Division, ABCNews.com, and 20th Century Fox. Suski senses that these companies ìall feel like they missed the boat on the first wave, the Internet, so they all want to be ready for this one.î

Corporate users will ultimately need to wait and see what practical applications emerge from this tempting technology. However, with all that money flowing in support of MPEG-4, opportunity is likely to be at the crest of that wave, and the boat is now boarding.

Copyright 2000-2001 Online, Inc.
213 Danbury Road, Wilton, Connecticut 06897-4007
203/761-1466, 800/248-8466
Fax 203/761-1444

<LAYER SRC="http://ad.doubleclick.net/adl/emedialive.com/home;sz=130x200;tile=2;ord=59?" width=130 height=200 visibility="hidden" on