Introduction to MPEG-4
- The MPEG Brain and Its Artifacts
- MPEG-4 in a Nutshell
- Architecture and Tools
- Next Generation of Portals
- Interactive Broadcast
This chapter provides a brief introduction to MPEG, an overview of the MPEG-4 standard, and a more detailed description of the architecture and tools of the MPEG-4 specification. The first topic casts a meme's eye on the MPEG organization and likens the MPEG body to a brain that produces standards as artifacts. The second topic provides a high-level view of the MPEG-4 standard by putting it in context and describing the fundamental principles behind the specification. The last topic goes one step deeper in the architecture of an MPEG-4 browser, thereby introducing the remaining chapters of this book.
The MPEG Brain and Its Artifacts
This section adopts a meme's point of view in describing MPEG, expecting that it will provide the reader with an interesting and accurate representation of what MPEG standards are and how they are produced. Formally, the Moving Picture Experts Group (MPEG) is a working group of ISO/IEC in charge of the development of international standards for compression, decompression, processing, and coded representation of moving pictures, audio, and their combination. Another way of defining MPEG is to take a meme's eye and see MPEG as a huge brain at the service of the memes that have a digital representation: the digital memes.
MPEG and Memes
An element of a culture that may be considered to be passed on by non-genetic means, especially imitation.
Further in her book, Blackmore expands the term to refer to memetic information in any of its many forms; including ideas, the brain structures that instantiate these ideas, the behaviors these brain structures produce, and their versions in books, recipes, maps and written music. As long as the information can be copied by a process we may call "imitation," then it counts as a meme.
Keeping the meme's definition in mind, one can make the following two remarks with regard to MPEG:
Some memes are transferred in audiovisual data. By providing a representation of this data that improves their quality and facilitates their replication, storage, and their broad diffusion, MPEG specifications are serving these so-called audiovisual memes.
Following the meme's point of view, the MPEG organization can therefore be seen as a structure at the service of the audiovisual memes.
Although these two remarks could be applied to most of the standardization body providing representation of digital data, MPEG, by the way it works and the nature of its activity, is a particularly good example, as we will see in the next sections.
The MPEG Brain
MPEG specifications themselves carry memes, but they are not the same as the memes carried in the digital representation of the media. They are contained in the description of the processes for decoding and reconstructing the audiovisual media.
A technological meme's chances of survival are given a real boost if it can enter a standard specification:
It is replicated in all terminals compliant to the standard.
It is assured a long lifetime, and it is exposed to a low risk of mutation since it is maintained by the ISO.
It stays together with other compelling memes, which makes them more attractive and forms with them what is called a memeplex.
One can therefore expect big fights during the standardization process: What is at stake for the memes is their replication or their death. The fighting process occurs during MPEG meetings. MPEG usually holds three meetings a year. These comprise plenary meetings where decisions are made and subgroup meetings where technical discussions occur. MPEG meetings are attended by some 300 experts from over 20 countries3. The selection pressure on the memes is increased since some of the memes are patented: Not only is the survival of the memes at stake, but also the interests of the companies that exert a strong financial pressure to have their intellectual property included in the standard.
MPEG, Memes, and Patents
Why is MPEG allowing patented memes in the standard? This seems to go against the broad diffusion of the technology.
The penalty of being patented can be compensated for by having active company supporters. Therefore, patented memes also have their chances since additional efforts are being made to promote them. In addition, having a memeplex with patented memes is beneficial to other nonpatented memes, since they benefit from the promotional efforts made by the supporters of the patented technology.
Therefore, patented memes may be damaging only to companies that have not been participating in the MPEG process or have no intellectual property in the standard. To avoid this, MPEG patent pools ensure that MPEG memes are made available to everyone at reasonable costs. These costs generally do not cover much more than the R&D investments and the patent pool logistic.
The fighting process is similar to what happens in our brains, where memesfor example, music, jokes, and poetryare fighting to be remembered and replicated. MPEG can therefore be likened to a huge brain, processing the memes contained in input documents, operating a selection on these memes, and outputting a pool of memes in the MPEG specifications. In such a competitive environment, memes use all the tricks they can to enter the specification in order to survive.
The selection of the memes in MPEG is done according to a set of design rules and a process named core experiment. The set of design rules includes
A tool-based approach: The MPEG memeplex is organized into small entities called tools that can be used independently. This approach allows the use of only part of the standard while guaranteeing that all the tools can work together. It therefore allows the creation of a memeplex momentum without putting any replicating burden on the individual memes.
Specification of the minimum: Only the tools that are needed to allow replication of the audiovisual memes (e.g., allow interoperability of the audiovisual data) are specified. In particular, only the decoding process is specified in MPEG.
One tool, one functionality: Options are a compromise between competing memes that permit both to enter the standard: "You let me in, I let you in." Still, they put an extra burden on the specification and compromise interoperability. MPEG has a strict rule to avoid duplication of functionality, and competing technologies are weeded out through a decision process illustrated in Figure 11.
Figure 11 The selection of the memes in MPEG is made according to a set of design rules and a process named core experiment.
The core experiment process consists of the following steps:
At a given MPEG meeting, the specification is frozen and described in a document called a verification model. "Mutations" of this specification are defined and are called core experiments. A core experiment defines the evaluation process of a given tool: data tests, description of the technology, and evaluation criteria.
Between meetings, the core experiment is carried out. In general, results are exchanged on mailing lists dedicated to that purpose. All the various mutations of the verification model are carried out in parallel and reported back to MPEG.
At the next MPEG meeting, the results of the experiments are gathered and decisions are made on the way the specification should evolve.
Finally, an important aspect of the MPEG process is that the MPEG memeplex is not documented only in plain Englishthat is, in a language only understandable by some human brains, with all its richness and approximation. MPEG also requires that the specification be written in a language understandable by virtually any programmable device (e.g., in C, C++, or Java). This requirement not only validates the specification one step further, it considerably improves the reproducibility and fidelity of the memes contained in the standard.
The MPEG brain provides for a selective environment for technological memes. After a strict evaluation, only a few of them are documented in textual and software standards, the artifacts of the MPEG activity. Even though some technological memes, such as those describing a decoding process, for example, get promoted through the MPEG process and therefore benefit from it, it is really the audiovisual memes that benefit most from the MPEG structure, as we will see from the existing MPEG standards and the ones under development.
So far, MPEG has produced MPEG-1, the standard for storage and retrieval of moving pictures and audio on storage media (approved November 1992), MPEG-2, the standard for digital television (approved November 1994), and MPEG-4, the standard for multimedia applications (approved October 1998). MPEG is now developing MPEG-7, the content representation standard for multimedia information search, filtering, management, and processing (approved September 2001), and MPEG-21, the multimedia framework.
MPEG-1 and MPEG-2 have provided the foundation for the digital representation of the audiovisual memes. Thanks to these standards, and in conjunction with the proliferation of networks to transport the encoded information as well as of terminals to display the content, audiovisual memes are ubiquitous. The main advantage of MPEG-2 over MPEG-1, from a meme's eye, is the possibility of having access to more human brains through TV broadcasting networks and terminals as well as through multilingual capabilities.
MPEG-4 extends the MPEG-1 and MPEG-2 models in three directions:
Object-based representation: MPEG-4 models audiovisual memes with a rich set of representations called audiovisual objects. They include high-level mathematical models such as curves and surfaces, not just compressed audiovisual samples. From a meme's eye, this means a richer palette of digital representations.
Interactive audiovisual scenes: Audiovisual objects can be grouped together in more complex structures called audiovisual scenes, to which a programmatic behavior can be attached. Reproducibility, as well as reusability of the memes, is improved since the data have an inherent structure. As the audiovisual content is more attractive, the impact of the memes on human brains is bigger.
Network abstraction: With DMIF (Delivery Multimedia Integration Framework), MPEG-4 provides an abstraction of the network so that MPEG-4 data can be authored once and carried on any transport mechanism with minimal adaptation efforts. With DMIF and the object-based scalability provided by MPEG-4, audiovisual memes can reach virtually any connected brains.
MPEG-7 is another breakthrough for audiovisual memes: MPEG-7's XML-based representation of the audiovisual data can reach the brains of the computers. MPEG-7 descriptions allow both low-level (e.g., color histograms) and high-level (e.g., data structure) reasoning on the audiovisual content. With these tools, audiovisual memes can be more easily searched, filtered, and retrieved by automated agents. Finally, with MPEG-21, the multimedia framework, MPEG is developing a complete ecosystem for richer and more-structured digital memes, not restricted only to audiovisual ones. MPEG-21 will allow these digital memes to circulate as freely as possible, according to the usage rights that human agents have attached to them.
Summarizing the meme's point of view of MPEG, we have seen that MPEG standards are evolving from MPEG-1 to MPEG-21 in such a way that the efficiency of the representation and replication of memes in human brains as well as in computer brains is constantly improved. We have also seen that the MPEG standard development process optimizes the memeplex formed by the standardized technology so that it best serves audiovisual memes.
The MPEG standardization body can therefore be seen as a structure working for the replication and diffusion of memes in general and audiovisual memes in particular. One may claim that this is not its sole purpose. Indeed, a rich economy is created around MPEG standards and could be seen as the primary goal of the standardization. Still, tastes and technologies change over time; sometimes quite rapidly. What will remain in any case are the digital representations of the memes, as if this were the only raison d’ être of MPEG, in the same way that what remains from ancient cultures are the artifacts that have endured through time.