Green gives a detailed description of Intel’s current ActionMedia software system and then motivates and describes the next-generation software system, called the Audio Video Kernel. The description, and presumably the design of the first-generation system, is based on the metaphor of a “super VCR”; the next-generation system is based on a new conceptual model, that of the digital video production studio.
Digital Video Interactive (DVI) is a system consisting of auxiliary processor hardware with special on-board algorithms for handling the massive data streams that constitute full motion video. The data are digitized, compressed, and stored on media such as hard disks or CDs, then decompressed and displayed.
The encoding approach emphasized is interframe coding, in which most frames are represented only by the changes in pixels from certain base frames, called intraframes. In addition to providing a way for video to be stored, DVI also provides for video to be transmitted across networks. This would allow sharing of archival resources, such as encyclopedias and films, and user-produced materials, such as student multimedia projects.
The current Actionmedia system works with IBM PS/2s or compatibles and the DOS operating system. The next-generation system is intended for a wider range of platforms and will take advantage of increasing power, both in the auxiliary system and in the main microprocessor.
The paper gives sound motivation for the original and the next-generation system, and the two conceptual designs are useful in understanding what is gained by the new approach. The paper includes many systems diagrams, and although the placement of the diagrams is not ideal, they are still useful. As someone with some familiarity with and interest in the field, I found the paper helpful for understanding design issues. It would be appropriate for a class in systems design.
The paper has several drawbacks when considered as material for Communications of the ACM, however. My assumption is that CACM papers are intended for people who are highly competent in computer science but not necessarily familiar with the requirements of a specific field. A side-bar describing the nature of compressed digital video should have been incorporated into the body of the text. More important, the paper suffers from what has been termed computer science present tense. I could not figure out the actual status of this “next-generation” AVK system. Is it more than a paper design? Is it operational? If so, are there any performance data to report? Can it handle the specific applications alluded to in the text (such as truck driver safety), and can those applications be described or at least characterized? More fundamentally, is it just the software or both the software and the hardware that is changing for the new system? Will the new function be observable when used by DOS machines? Intel’s product development has been discussed in the trade press, and it would be appropriate to show the reader how the technical evolution described here correlates with the product direction. Also, some explanation of the relationship, technical and sociological or political, among the different standards and interfaces (including JPEG, MCI, and HDTV) would have been beneficial.