It's only software

Offering high-quality video files with today’s demands for multicasting and alternative distribution media requires extra attention to encoding and decoding methods
Publish date:
Social count:

For seasoned veterans of the broadcast industry, the traditional concept of a video encoder has been — and in many cases still is — a piece of hardware that takes in a video signal and encodes it for emission via the transmitter.

In the analog world, this typically involves taking analog component signals from a camera (RGB) or a piece of component production gear (Y, R-Y, B-Y) and compressing the wideband analog components to fit in a 6MHz NTSC broadcast channel. NTSC encoders are typically built into products so that the analog baseband outputs can be routed throughout the plant and mixed with other NTSC signals.

As described in October's Download column, the transition to DTV has been accompanied by new ways to think about video encoding and the proper point(s) in the signal flow of a DTV facility where encoding takes place. Unlike baseband analog signals, compressed digital signals cannot be mixed. They must first be decoded to digital component baseband signals.

For most DTV broadcasters, a real-time, hardware-based MPEG encoder sits between the output of the master control switcher and the input to the DTV transmitter. A far smaller group of broadcasters is moving to systems that splice emission-quality MPEG-2 streams to manage traditional master control tasks.

But software encoders are proliferating outside of the world of DTV broadcasting. Chances are that they are going to play a key role in the future of program delivery, much like they do today in the production of those tiny plastic DVDs that are filled with movies and episodic television programming (without the commercials).

If you are in the business of delivering program files to consumers, software encoders may offer important advantages over real-time hardware. Even if you are in the business of delivering real-time program streams, if they are not live events, there may still be an advantage to using a software encoder and then playing out an optimized file.

Time vs. quality

The most important distinctions between software and hardware encoders are time and image quality. The reality is that digital video encoders are only software — algorithms that run either on dedicated chips optimized for real-time applications or general-purpose CPUs.

To make things even more confusing, most digital compression standards do not specify how an encoder should work; they define the syntax of a properly encoded bit stream so that a compliant decoder can turn the stream back into digital component video for display. Product developers are free to do anything they want, in terms of encoder design, as long as their products produce compliant bit streams.

The problem with real-time encoding is time. A new field or frame must be encoded in 1/24th of a second or less, depending on the field or frame rate of the digital source. The time-consuming part of most compression algorithms are the routines used to build predictions of what an image should look like, using already decoded frames that occurred both before and after the predicted frame.

MPEG-2 uses block-matching routines, which can either be done fast or done well. The new H.264 compression algorithm adds significant complexity to these prediction routines and allows for more anchor frames to be used to build the predictions. When time is a factor, these routines may be truncated, but the predictions will not be as good as they could be. The net result is that it will take more bits to send the differences between the prediction and the actual frame. When the channel bandwidth is fixed, this often results in the generation of compression artifacts.

On the other hand, with a software encoder, the prediction routines take as long as is necessary to produce a desirable result. And, if the quality is still not adequate, a compressionist can use additional tricks to improve the delivered image quality, such as forcing I frames at scene transitions or applying pre-filters to reduce the information content so the encoder can produce high-quality images at the available bit rate. Many hardware encoders use pre-filters to reduce information content as well.

When someone goes to the trouble to do the best possible job of encoding, it would seem appropriate to allow the consumer to realize the benefits of that work. That's exactly what happens when you watch a movie on a DVD.

No longer just an MP3 player, the new iPod from Apple allows users to watch home movies, music videos and television shows on a 2.5in color screen. The iTunes store lets users download five popular shows, such as “Lost” and “Desperate Housewives,” as well as six shorts from Pixar Animation Studios.

But broadcasters who place the emission encoder at the end of the operations chain can undo all of this extra effort. If a source is encoded with high quality, then decoded for mixing in the master control switcher, the resulting quality may be severely reduced by the real-time encoder at the output of the switcher. Today, all of the commercial networks — except FOX — send their affiliates contribution-quality video, typically encoded at 45Mb/s for HD and some fraction of that for SD. Because of the higher bit rate, contribution-quality video is typically quite good, but that quality is reduced when it is encoded at emission bit rates.

Clearly, there are benefits to proper optimization of digital compression, but these benefits can only be realized when those optimized bits are delivered to the consumer. This is where emerging business models come into play.

Where do you want your content to go today?

The era when a broadcaster could focus all of his or her resources and attention on the delivery of one program stream is drawing to an end. Today we often hear that the success of broadcast DTV will depend on multicasts. Increasingly, broadcasters are looking to deliver multiple programs in their DTV multiplex, not to mention the need to make some content available via alternative distribution media, such as a station's Web site or a cell phone-based video service.

DISH Networks entered the portable video player arena with its PocketDISH, available in three models. The units allow users to transfer programs from their DISH Network DVRs for viewing on the go. Like the iPod, users can also store music and photos on the units.

As described in last month's column, consumers are becoming program directors. With electronic program guides and personal video recorders, viewers can now capture programs and watch them at any time. If a program is being delivered to a local cache, it would seem appropriate to make certain that the encoded bits are properly optimized for the delivery channel at the best quality level possible given bandwidth constraints. This may mean one form of encoding for video is sent to mobile devices, such as cell phones, a higher-quality level is for Internet downloads, and the highest-quality level is for HD programming delivered via DTV broadcasts. Software encoders are often the best choice when it is necessary to generate multiple versions of your content.

Apple and DISH Networks have entered the video download business and introduced portable video players. The new iPod uses H.264 and MPEG-4 compression for video files that are sold via the Web-based iTunes store. The store now offers more than 2000 music videos for download and several hit ABC series, which are available the day after broadcast. The PocketDISH players work with a standard DishPVR, allowing files to be transferred to the portable players for use anywhere, anytime.

Equally important, most of the new opportunities to expand the reach of your content are being driven relentlessly by the continuous evolution of software encoders and decoders. When Apple or Microsoft releases a new version of QuickTime or Windows Media, it is not uncommon for several hundred million decoders to be downloaded in a month.

Apple continues to have a loyal following of video content creators for both the Mac and PC platforms, in large part because of the QuickTime digital media architecture. This architecture supports a long list of proprietary and standards-based codecs for audio, video and still images, and it allows manufacturers to extend the capabilities by providing their own plug-in components. (See “Web links.”) The QuickTime container format forms the basis for the MPEG-4 file format, enabling sophisticated local video composition capabilities when the entire MPEG-4 standard is used.

Microsoft has taken a proprietary route with Windows Media technologies (see “Web links.”) and has succeeded in getting its VC-1 compression technology adopted for the next generation of HD DVD products and many IPTV deployments.

The bottom line is that software encoding and decoding is going to play an important role in the future of television. In a world where people are moving from surfing 500 channels of TV to searching for and caching the programs they want, delivering high-quality video files optimized for a wide range of playback devices may be critical for survival.

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

Web links

“Viewers: The new program directors”

Apple QuickTime

H.264 FAQs

MPEG-4 products and services

Microsoft VC-1

Send questions and comments