'MP4' Format to allow interchange, streaming of MPEG-4 Multimedia Content

San Jose, CA, February 1998 - MPEG has begun work on its MP4 file format (previously known as the "Intermedia Format"), the format that can be used to exchange MPEG-4 content, and that supports streaming MPEG-4 content over a variety of networks and bitrates. The design of the file format has started from responses to a Call for Proposals. Proposals were submitted by the EBU/SMPTE Task Force, THOMSON multimedia, AT&T Research, Columbia University, Microsoft, Intel, Apple Computer, IBM Corporation, Netscape Communications Corporation, Oracle Corporation, Silicon Graphics, Inc., Sun Microsystems and Avid Technology Inc. Following the usual MPEG work method, the submissions were thoroughly analyzed, and while the development work has started from one of the submissions, all options are open for improvement and the only a priori requirements taken into account are those that MPEG has drafted at its previous meeting. MPEG is very pleased to see that new MPEG members drawn from these companies have rapidly assimilated the MPEG spirit and are collaborating in the development of a fundamental component of the multimedia world that MPEG-4 creates.

Work has also started on the "Adaptive Audiovisual Session" (AAVS), which will provide a set of Application-Program Interfaces to give MPEG-4 Java programs access to e.g. terminal resources, input devices, application programs etc. AAVS will provide a valuable extension to MPEG-4 Systems, giving content and application developers flexible and powerful tools to create multimedia content. In addition to giving them advanced capabilities for programming ‘content behavior’, it allows them to, for instance, determine how content should scale to the decoding device. Also, AAVS supports both local and remote user interactivity.

First steps were taken to build protection of intellectual property into the architectural framework of version 2 of the MPEG-4 standard. The goal of the effort is to give content producers and owners the technical possibilities to prevent their content from illicit access and reproduction.

MPEG has taken up efficient representation of Body Animation (to complement the Facial Animation activities) and 3D models.

Although a number of new activities were kicked-off, MPEG’s 42nd meeting in San Jose, CA, was mainly dedicated to checking the details of the Committee Drafts (CDs) of the Audio, Video, Systems and DMIF parts of the MPEG-4 Standard, in anticipation of the comments to these CD’s. These will accompany the official ballot votes that will be received and processed during the March MPEG meeting in Tokyo. In the light of this work, many bitstreams have been exchanged to verify the soundness of the specifications.

A special syntax in the MPEG-4 standard provides the decoder with an early indication of decoder resource requirements (processing and memory), to permit a decoder to optimally load balance its available resources and so avoid unacceptable degradation of decoding quality. This is called ‘Computational Graceful Degradation’.

Ongoing tests on the quality of the new Advance Audio Coding’ (AAC) part of the MPEG-2 standard, show that the AAC Main and Low Complexity profiles are able to demonstrate full broadcast quality at 128 kbit/s for stereo, i.e. approximately half the bitrate of that needed by the earlier MPEG-1 Layer II codec, which was included in these tests as a direct comparison.

A first draft was issued of the ‘Proposal Package Description’ (PPD) for MPEG-7, a new standard to allow locating multimedia content. The PPD describes the expected content of a proposal for MPEG-7 technology. The Final Call for Proposals will be issued at the end of this year.

The meeting in San Jose was held in San Jose of the US National Body NCITS L3.1 and organized by the Interactive Technology Center of Lockheed Martin Telecommunications Company.

MPEG-4 Version 1 will become ‘International Standard’ in December 1998, while (the backward compatible) Version 2 will follow one year later.

Details

The remainder of this press release contains the details of the meeting, organized according to the subgroups in which MPEG is organized.

Systems: Extending MPEG-4’s multimedia capabilities

Next to the issues mentioned in the opening of this Press Release, work is ongoing on the so-called ‘Advanced BIFS’ (BIFS: BInary Fromat for Scene description). A first set of new BIFS nodes have been identified, giving MPEG-4 functionality existing in VRML and not yet supported by MPEG (ex : PROTO, scripts, …). Also new MPEG-4 specific nodes (e.g. advanced spatialization of sound reproduction, multi-user interaction) have been specified

Video: Rigid testing of the tools

Anticipating the votes for MPEG-4 CD in March, it was decided that Gray Scale shape coding is ready for inclusion in MPEG-4 Version 1, in answer to a wish expressed by many broadcasters in MPEG-4.

The emphasis in the work of the video subgroup was on verification of the Visual part of the MPEG-4 Standard. This consists of:

• bitstream exchanges, in which all the operation of all coding tools and syntax of MPEG-4 Visual were verified, and
• implementation of all tools in two common software platforms, to ensure correct specification of the standard. The source code of these two implementations will form part of the final standard.

Comments were also received from a number of national bodies on the Committee Draft released in November 1997. Based on these comments and other verification, work was commenced on the Final Committee Draft of MPEG-4, to be released at the Tokyo meeting in March 1998.

Audio: AAC shows superior performance

Continuing its work aimed at proving the viability of the MPEG Standards, MPEG members reported detailed tests on Stereo Audio MPEG-2 ‘Advanced Audio Coding’ (AAC) coders working to the Main, Low-complexity and Scalable Sample Rate Profiles. These represented the culmination of a period of International cooperation to demonstrate the achievements of the AAC developers.

The tests were conducted according to the most stringent requirements of the ITU-R Recommendation BS-1116 test methodology at the Research Laboratories of NHK in Tokyo. In total, eight combinations of bitrate and codec were assessed using 10 specifically critical programme items and 31 listeners. Rigorous statistical procedures were invoked in order to ensure that only fully justifiable conclusions were drawn from the test data. The full report of these tests can be downloaded from the MPEG Home page (see below) and shows that the AAC main and low complexity profiles are able to demonstrate full broadcast quality at 128 kbit/s for stereo, i.e. approximately half the bitrate of that needed by the earlier MPEG-1 Layer II codec, included in these tests as a direct comparison.

In 1993, MPEG had set itself the target of achieving high quality audio at a bitrate of 64 kbit/s per channel. This has now been successfully achieved, representing the successful culmination of a significant amount of work, with tremendous international collaboration.

Synthetic Video: 3D models and body animation in MPEG-4 V.2

Significant international participation is rallying for MPEG-4 Version 2 work in the areas of 3D Model Coding and Body Animation. These efforts are expected to dramatically increase the efficiency in downloading 3D models, to provide complexity scalability in the shape and appearance of such models, and to elaborate the coding of Body Animation to complement Face Animation. Test data sets for these efforts are planned to draw on VRML format models. Results of this coding work are expected to allow models like the VRML format invoked in MPEG-4 to be compressed and integrated within MPEG-4 streams.

The 3D Model Coding effort draws important competing international contributions to four areas of model compression and scalability of 3D meshes and their shape and appearance: compression of connectivity, compression of geometry, progressive coding, and surface properties (e.g. color, texture mapping, surface normals). Experiments are now carried out to select and combine the best elements.

The Body Animation coding now includes focus on animation parameter compression, specialized animation parameterization for the hands, quantization error effects in Body Animation coding, and interpretation of parameters for Body definition and calibration.

Implementation Studies: MPEG-4 suitable for low complexity systems

An analysis of the Video Decoder complexity indicates that low cost systems and software-only implementations for MPEG-4 are achievable, enabling early adoption of MPEG-4 technology in products and systems.

Mechanisms for constraining the decoding complexity of MPEG-4 bit-streams were identified for natural and synthetic. Further progress was also made with regard to Computer Graceful Degradation (CGD) in the areas of video and more recently synthetic media. CGD information provided in the bit-stream provides an early indication of decoder resource requirements (processing and memory), so as to permit a decoder to optimally load balance its available resources and so avoid unacceptable degradation of decoding quality.

DMIF: preserving QoS across networks

A demonstration for the integrated software using MPEG-4 IP transport, MPEG-4 broadcast and local file storage, is targeted for the 44th MPEG meeting in Dublin, which takes place from 6-10 July 1998.

DMIF Version 2 contains now an architecture for DMIF Interworking Units (IWUs) which preserve the QoS across heterogeneous networks, for example consisting of HFC and ATM. This allows the management of the resources on the separate networks to achieve an end-to-end media based QoS specified at the DMIF Application Interface (DAI).

Background information

Future MPEG meetings will be held in Tokyo, JP (March '98), Dublin, IE (July '98), Princeton, NJ, US (Oct. '98), Eilat, IL (Dec. '98) and Korea (Mar. '99). An MPEG-7 seminar with invited speakers will be organized during the meeting in Tokyo on Wednesday 18 March. This seminar is free of charge, and (unlike the normal MPEG meetings) open to non MPEG delegates as well.

For further information about MPEG, please contact:

Dr. Leonardo Chiariglione, (Convenor of mpeg)

CSELT

Via G. Reiss Romoli, 274

10148 Torino, ITALY

Tel.: +39 11 228 6120; Fax: +39 11 228 6299

Email: leonardo.chiariglione@cselt.it

This press release and a wealth of other MPEG-related information can be found on the MPEG homepage:

http://www.cselt.it/mpeg

The MPEG homepage has links to other MPEG pages, that are maintained by some of the subgroups. It also contains links to public documents, that are freely available for download to non-MPEG members.

Also, a web page specific to MPEG Audio matters has been established. In the four weeks since it was made public it is already proving to be very popular. The URL of the MPEG Audio Web Page is:
http://www.tnt.uni-hannover.de/project/mpeg/audio/

Journalists that wish to receive MPEG Press Releases automatically can contact:

Rob Koenen
KPN Research
tel. +31 70 332 5310; fax +31 70 332 5567
Email: r.h.koenen@research.kpn.com