Customer Reviews

The MPEG-4 Book

5 star:		(0)
4 star:		(0)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

Because large parts of MPEG-4 are still unimplemented, and therefore were unimplemented when this book was published, it is mainly about how one would implement MPEG-4 according to the standard along with pretty good details on different parts of the technology. Unlike MPEG-2 where the killer application was digital television, MPEG-4 does not target a major and exclusive killer application but opens many new frontiers. The first two chapters are mainly an overview of the standard - its origins, its objectives, and its possibilities.

All components of an MPEG-4 presentation are discovered recursively through object descriptors (ODs). The details of this approach are explained in chapter 3. SDL is a convenient shorthand to express bitstream syntax of inherently object-oriented structures in a C++ kind of way, and is explained in detail in the context of MPEG-4 with several practical examples.

Chapter 4 on the "BIFS", binary format for scenes, is another practical chapter. This chapter presents the principles of MPEG-4 scene description, which is BIFS. BIFS is the scene description format designed by the MPEG Systems subgroup [MPEG4-1] to complement the object descriptor framework presented in Chapter 3. Whereas the OD framework defines the objects and their characteristics, BIFS defines how the objects are combined together for presentation. Numerous example are shown along with the scenes they generate to help explain the concept. You won't be able to help seeing the similarities between VRML and MPEG-4 scene descriptions.

Chapter 5 examines MPEG-J. MPEG-4 terminals may vary from high-quality TV set-top boxes to wireless and handheld devices. The computational capabilities and the network resources available at these terminals differ considerably. Without MPEG-J, these wide variations in resources would make it very hard to achieve the dream of create once, run anywhere. Thus, the objective of MPEG-J is to add to the presentation engine an application engine that can associate complex programmatic behavior with the presentation based on user input and terminal conditions. Although this chapter does a very good job of documenting what the MPEG-J API's shall contain and do according to the standard, the fact is that nobody has implemented them. This leaves the reader with the false feeling that this is something that is available.

Chapter 6 is about XMT, which is a textual representation of MPEG-4 system structures. These MPEG-4 systems structures are represented as XML elements, with attributes of the element containing its values. This can all become very complex, with considerable overlap between SMIL, MPEG-7, XMT, and X3D. This chapter does a good job of sorting out the difference and interoperability of all of the standards through diagrams and examples.

Chapter 7, Transporting and Storing MPEG-4 Content, details the delivery abstractions defined by DMIF (Delivery Multimedia Integration Framework), and then explains the FlexMux tool, which is a simple syntax to interleave data from various streams. This is followed by in-depth coverage of the MPEG-4 file format as well as the specification of MPEG-4 over MPEG-2 Systems transport. Finally, a summary of the ongoing work to define the transport of MPEG-4 over IP is provided.

From this point on in the book, everything is very theoretical. There are good clear examples of what speech coding, natural video coding, and general audio coding all are in general and according to the standard, complete with mathematics and good instructive figures. However, if you are coming to this book to find out who has implemented this part of the standard and how to use MPEG-4 audio coding "out of the box", in a practical sense, you will be disappointed. The only exception to this is chapter 12, on synthetic audio, where the AudioBIFS and Advanced AudioBIFS are introduced and explained via some very clear examples. Again, you are going to see big similarities between AudioBIFS and VRML.

In summary, I would say that if you want a good understanding of the MPEG-4 standard and the technologies behind it, you will not be disappointed. If you are coming to this book as an application programmer who wants to insert MPEG-4 capabilities into a program you are writing in much the same way that you download and use the Java3D API, you will leave unsatisfied, because much of this standard is yet to be implemented. Also, much has changed over the last four years so a second edition of this book pointing out what has actually been implemented during this time would be helpful. Given the book's age and lack of implementation details on the trickier parts of MPEG-4, I wouldn't recommend paying more than the discounted used-book price for it.