MPEG coding systems

Things are not always what they seem. We routinely pass pictures around represented by only a couple of percent of the original picture data. It is a fortunate fact that we can use representations of reality to reconstruct what seems to be reality.

The Leitch NEWSFlash-II nonlinear editor can edit multiple compression formats and render the output to the compression format of choice.

MPEG unleashed an industry that has improved the quality and reduced the cost of transmission, and has become the basis of the digital terrestrial television system worldwide.

Today, commercial implementations of MPEG can be found in DTV, consumer DVD, camcorders and hard disk recorders for consumer and professional applications, and transmission systems for satellite and terrestrial distribution. The range of quality and extent of features supported is similarily stratified.

SDTV MPEG hardware offers a wide range of features. (HDTV MPEG encoders are in use primarily in North America.) At the high end are efficient encoders intended to wring out the last possible bit of redundant information in the content. They may be installed in facilities using statmuxing to further increase efficiency. Full-featured systems often include interfaces for SMPTE 259M and AES audio, and embedded audio. They can accommodate multiple audio channels, which may be useful for multiple language requirements.

These “high end” encoders sometimes use a “two-pass” process to improve the efficiency. This works by doing a first encoding pass and looking at where the content was not challenging and bandwidth was unused. The uncompressed input signal is then passed through the encode signal again, using statistics gained from the first encode cycle to tell the second pass where it improved the picture. The results can be quite dramatic at low bit rates.

The same technique is used when coding for DVD release, though it may well be a two-pass, operator-assisted process. In a suite similar to a film transfer environment, a “compressionist” optimizes the bit rate for each scene.

The middle ground is filled with encoders optimized for program backhaul that work without intervention. Settings are typically determined for average content. One technique to optimize the results is to use preprocessing on the input signal to allow the encoder to act more efficiently. Preprocessing might include noise reduction and high-quality decoding for composite signals. The cleaner the video is at the input, the lower the effective bit rate can be for the same quality at the decoder.

These mid-grade encoders are excellent for SNG and ENG purposes. They often offer optional internal RF modulators, and outputs for connection to terrestrial circuits as well (DS3, ATM, SONET, etc.). Packaging might include weatherproof cases and small size for use with wireless cameras.

The third tier of encoders is intended for very low cost and modest quality, including encoders built on PCI bus cards and small stand-alone encoders. These can be useful for low-bit-rate applications like remote monitoring, education, and streaming in corporate applications. They may well see application in the future in “barker channels” in a DTV multiplex where the moving video is restricted to a small portion of the screen, and text and graphics with no moving content surround. The restriction of moving video to one portion of the screen makes coding at low bit rate and low cost easily achievable.

A short word on the emerging MPEG-4 products is in order. Many applications we use today are closed loop, and compatibility with large universes of decoders is not needed. That is where the genius of MPEG-2 has provided huge benefits. However, when one seeks the most efficient compression possible, MPEG-4 should be considered if the application is narrow and controllable. Bit rates can be dramatically lower, or quality raised for the same modest bit budget. Look for MPEG-4 to seriously penetrate the professional market in the next year, including HDTV backhaul applications.