Starting from the idea of compressing digital video to fit it in the low throughput of 1.4 Mbit/s of a CD, MPEG had to develop newer areas of expertise to provide standard component technologies to build complete solutions. The size of the group (some 300 participants from early on, 500 experts at the time of the High Efficiency Video Coding (HEVC) standard development and now 400 participants) and the span of technologies considered has prompted the establishment of a subgroup structure so that work could proceed more expeditiously.
Following the order of the MPEG workflow, two examples of which (MPEG-1 and MPEG-2) have already been provided, the current subgroups are:
- Requirements (RQ) is produces requirements for standards, both new or already under development.
- Systems (SY) develops standards for the coding of the combination of individually coded audio and video, including related information so that the resulting combination can be used by an application.
- Video (VD) develops standards for coded representation of moving pictures of natural origin.
- Video Coding (VC) develops the High Efficiency Video Coding (HEVC) standard (joint with ITU-T).
- 3D Video (3V) develops 3D Video extensions for AVC and HEVC (joint with ITU-T).
- Audio (AU) develops standards for coded representation of audio information.
- 3D Graphics Coding (3D) develops standards for coded representation of audio and moving images of synthetic origin.
- Test (TS) develops methods for subjectively evaluating, tests and verifies the quality of coded audio and moving picture.
- Convenor Advisory (CA) advises on general matters related to the group, specifically schedule and organisation of meetings.
MPEG has been one of the first groups to exploit the power of the Internet not just to exchange emails and set up discussion groups, but also to create a repository where members upload and download input documents, so that all other members can download the submissions, study them and discuss the issues raised on email reflectors even before a meeting start.
This facility has enabled the group to increase its productivity to levels unthinkable before, when submissions could only be considered after the beginning of the meeting, actually after the host had made hundreds of copies for each of the input documents. Currently several hundred input documents are uploaded at every meeting as input contributions (during the development of HEVC more than 1,000). The upload must be made at least 6 daysbefore the meeting starts.
The idea to put in place this new method of work started at the New York meeting in July 1993. Managing input documents, MPEG’s most valuable asset, had been a nightmare since early times. Many participants were already making good use of email, but not all of them had access to it and, for those who had, adding a Word document as an attachment to an email would often lead to unpredictable results spurring a ping-pong of well-intentioned emails designed to solve the problem of “format incompatibility” (a.k.a. lack of standards). For a long time the safest way to have a document considered before the meeting was to send preliminary copies of contributions by fax to members who were assumed to be interested in it and then bring an original to the meeting. The host would then make 200-300 copies of each of the 100-200 contributions. Assuming an average of 8 pages per contribution and adding the copies of the documents that were produced at the meeting, the host had to often produce more than half a million pages of photocopies, with the record of 1 million copies first reached in November 1993 in Seoul, which included output documents on paper and on transparencies.
The other problem was document distribution inside the meeting. The British Standard Institute, who hosted the November 1992 meeting in London, discovered at its expense that the standard “pigeon holes” method, besides being very costly to administer, simply did not work for MPEG. The only method that seemed to work was making piles of copies for each of the documents from where delegates would draw those of their interest. Jacques Guichard, former Director of Human Interactions at France Telecom R&D, and a long-time attendant at the COST 211 and CCITT meetings, seeing the rush of delegates to get a copy of an important document, once remarked that the level of civilisation in MPEG was lower than CCITT’s. Maybe so, but it is likely a consequence of the fact that MPEG is a large, totally self-supported organisation (besides being populated with unruly members). And so it continues to be to this day.