Choosing compression formats

Both AVC and WM9 have been shown to achieve comparable quality to MPEG-2 at half the bit rate
Image placeholder title

The EnvivioTV MPEG-4 player for set-top boxes, PCs and mobile devices is available as a plug-in for QuickTime, RealOne, Real Player and Windows Media player products.

Compression technology has evolved to the point that “codec wars” rarely get press, and producers can choose tools for their features and usability rather than being concerned about quality constraints. Consumers have benefited by the proliferation of streaming media content in their format of choice and a higher quality and quantity of content on DVDs.

Streaming media formats

This is good news all around, but if you are a producer, there is still the issue of how many formats you can realistically support, or whether you can limit your production to standards such as MPEG in today's market. Here's a look at how MPEG stacks up to other proprietary compression formats for producing content online, on DVDs and in high definition.

The streaming media industry has grown consistently over the last few years, with news and entertainment leading the way for online content producers. For playback on the desktop, MPEG-4 competes with proprietary codecs from Sorenson, Real and Microsoft, among others, in delivering streaming media. Apple's QuickTime and Macromedia's Flash are also streaming media formats that act as containers for compressed video playback on the desktop.

With the adoption of the AVC/H.264 codec into the MPEG-4 standard, its competitiveness in the streaming market has risen greatly. High-quality streams can be achieved at broadband rates over 500kb/s in any format, but still, many Internet connections cannot sustain such a bandwidth as a requirement for streaming audio and video files. RealVideo and Windows Media have an advantage in their approach to scalability, whereby a single encoded file can contain multiple bit rates. When streamed from their perspective servers, the bit rate can be automatically adjusted to fit varying network conditions. MPEG-4 has several specifications for scalability, such as its Fine Grain Scalability (FGS) profile, but there is little in the way of implementation so far. While broadband to the home continues to build out, there will be a need for such scalability for the foreseeable future.

MPEG-4 streams can be played on any of the three major players on the desktop using plug-ins available from Envivio. They also provide the most advanced toolset for creating MPEG-4 content. Apple continues to support the standard, allowing encoding and playback of the Simple Profile in QuickTime. It has announced support for the AVC codec this year, but this lag time between adoption of new technology and availability of tools is often used as an argument against relying on the standard.

With MPEG-4’s efficient object-based encoding, video images can be separated into components, such as a background and moving foreground objects, referred to as sprites.

MPEG-4 separates itself from the crowded field of streaming media codecs with its approach to creating and delivering rich media content. Based on the QuickTime file format, the standard supports many different media types, or tracks, such as text, sprites, textures, synthetic music and images, which can be packaged along with the standard audio and video content. Its approach to compression is then object-based, allowing for higher levels of efficiency and interest in the content. For instance, text need not be compressed as an embedded element of the video, but it can be sent alongside it and rendered in the client's viewing device. Different audio tracks can be included in a file, allowing the user options such as the type of background music or the language they wish to listen to. This is one way that MPEG-4 addresses the scalability issue. Because varying levels of complexity can be encoded into a file, and with intelligent server technology, clients can receive only those elements that fit within their network bandwidth.

New formats for prime time

Very low bit-rate content is another growing sector in the streaming marketplace, with many manufacturers committing MPEG-4 technology to their hardware devices and others also supporting RealVideo and Window's Media. The object-based format specified in MPEG-4 and its ability to encode synthetic music and images with texture maps could prove effective in delivering training videos to field technicians at bit rates below 10kb/s. Today we see little in the way of real implementations for this object-based format, but great potential exists in the standard that competes in the streaming media market, a format that may eventually help to define new applications in the future.

Standards play a more important role in the digital television and set-top appliance technology than in the flexible PC platforms that playback streaming media. While MPEG-2 is the accepted standard for DTV today, the greater compression efficiency of MPEG-4 and WM9 bode well for entry into this market. Last year, Microsoft made a bid to SMPTE for acceptance of its VC-9 codec, a subset of WM9, as a standard. As a joint development project, AVC is backed by the ISO and the MPEG committee as a standard compression technology.

Advances in compression efficiency translate into higher channel capacity within a given bandwidth allocation, which is particularly promising for the roll-out of high-definition programming. Both AVC and WM9 have been shown to achieve comparable quality to MPEG-2 at half the bit rate. Not only do today's satellite and cable operators benefit from increased channel capacity, but more efficient compression opens new markets for transmission over DSL, with full resolution, standard-definition television being encoded below 1Mb/s. It seems reasonable to assume one or both of these formats may someday replace MPEG-2 to transmit television to the home, especially for VOD services and as a recording format when such devices are available.

MPEG-4's efficient object-based encoding that we are just beginning to see on the Internet could spawn new television services as well. As shown in the image above, video images can be separated into components, such as a background and moving foreground objects, referred to as sprites. The panoramic view of the background would be encoded separately from the golfer and the ball on the course. Not only does this offer more efficient coding of motion information, but in conjunction with two-way communication networks to the home, users could also choose different viewing angles, just as they do on some DVDs today.

The DVD Forum also has given preliminary approval to both MPEG-4's AVC and Microsoft's WM9 (VC-9) codecs for their HD DVD specification. Final approval is pending licensing issues that have muddied the waters for MPEG-4 adoption and raised questions about the proprietary nature of WM9 this past year. One can imagine how the object-based encoding could fit well with the components of the DVD specification, including text tracks, various audio tracks and a means of efficiently encoding for different viewing angles as well.

Clearly, technology developments need to advance, production tools need to evolve and market demand needs to rise before all the interactive features and efficiencies of MPEG-4 can be realized. For basic rectangular video and audio content played back on the desktop or the set-top appliance, several contenders remain in the field, with adoption of a standard possibly being left as a viewer's choice in the end.

Barb Roeder is president of BarbWired, a technology consultancy specializing in the formatting and delivery of digital media.