Not long ago, multimedia was being described as a "zero-billion-dollar industry." But then the gold rush arrived in full force over last year's holiday season and has now kicked into overdrive. This is true across the nation, but the stampede seems particularly evident in California, especially in San Francisco, where you'll run into store clerks and auto mechanics--including grayhairs with WWII-vintage tattoos, not just the twentysomethings with nose rings who live and work in "Multimedia Gulch"--busily authoring a multimedia title or two. In Hollywood, the computer industry is doing lunch en masse with the entertainment industry: You may be a successful film star or filmmaker, but you're nobody unless you've signed a contract to produce a multimedia title with a computer-game publisher. It's a sign of the times when the likes of ex-CIA director William Colby are designing an interactive spy experience. Moving from the devil to the deep blue sky, IBM has announced an exclusive licensing agreement with the Vatican to digitize the vast library of illuminated manuscripts and works of art spanning centuries.
Like those who sold eggs to goldminers, suppliers abound to feed those who want information and know-how via tradeshows, conferences, and magazines. InterMedia kicked things off in March, followed by the QuickTime Multimedia Conference in early April. Then it was the NewMedia show in mid-April, Silicon Interactive in early May, and finally VisiComm NewMedia in early June. Dave Bunnell's New Media was one of the early entrants into the field of developer-oriented magazines and is now almost venerable, while Doug Millison's Morph's Outpost On the Digital Frontier (which sometimes calls itself "the Dr. Dobb's of multimedia") is considered the young turk. In consumer-oriented magazines, IDG has three overlapping titles: MultiMedia World, Electronic Entertainment, and Desktop Video World (recently renamed Digital Video). Ziff-Davis is establishing a new division that will produce magazines, multimedia titles, and online publications. And for the financiers, there is The Red Herring, a pricey magazine ($15 per issue) that advertises: "Before you invest in multimedia, invest in the Red Herring, the only magazine that covers the convergence of entertainment, computers, and communications from a strategic business perspective." While selling eggs to goldminers may be lucrative, it seems you can charge even more by selling chickens to those who sell the eggs.
In this article, we'll look at the technology aspects of multimedia and consider the implications for in-the-trenches programmers. But first, some rough measure of the scale of this gold rush. In 1993, Apple sold one million CD-ROM drives, a twenty-fold increase over the previous year. Dataquest expects the installed base of CD-ROM drives to grow from four million in 1993 to six million by the end of this year. According to the Wall Street Journal, 80 percent of these are used for multimedia applications, rather than retrieval of textual data. Apple says that 80 percent of its new PowerPC Macs ship with CD-ROM drives installed. The numbers from Microsoft tell a similar, albeit slightly inconsistent, story. According to Microsoft's Paul Osborne, product manager of Video for Windows, 12 million Windows-capable PCs will be shipped this year, of which "40% are multimedia machines," resulting in "400,000 new customers per month" for multimedia titles. (What's not clear is whether the definition of "multimedia machine" includes a CD-ROM drive; if so, this number appears inconsistent with the numbers from Apple and Dataquest.) The majority (60 percent, says Microsoft) of these multimedia machines are ending up in the home, rather than in the corporate office, and this lends emphasis to the entertainment portion of the market. So for now, the equation is simple: CD-ROM means multimedia, which means interactive entertainment.
In theory, you don't have to be a programmer to author a multimedia title, because high-level tools such as MacroMind Director and Adobe Premiere are supposed to enable anyone to create an interactive work--in the same way that tools like Adobe Illustrator allow any visual artist to create desktop-published graphics. But, in practice, most successful teams have a programmer on board, if not driving the project.
One who has achieved some success is Farshid Almassizadeh. At 24, Farshid is still officially a student at UCSD, not quite dropped out, a double major in computer science and "lighting and design." But he and his cofounders at Presto Studios released the highly regarded CD-ROM-based game, "The JourneyMan Project," in January 1992, now within striking distance of selling 100,000 units. The founders of Presto Studios bootstrapped their first title by working other jobs, but now have the resources to devote 11 more-than-full-time staff members to producing the sequel, which is estimated to require about 30,000 work-hours of development. A significant portion of this effort, about 10,000 hours, is programmer time. Farshid is jazzed about the upcoming release of MacroMind Director 4.0, because it now includes a compiler for the Lingo language (not just an interpreter) as well as powerful new constructs like heterogeneous lists. Along with increased control over memory management, these improvements will result in better performance and greater flexibility compared to previous versions of Director.
You can see parallels to the "Fire in the Valley" era of Silicon Valley, those golden days when it seemed that any dropout could start a successful company in a garage. The only difference is now you must do it in a loft. These opportunities are refreshing during a time in which the major software firms are following the path of consolidation and merger blazed by the auto companies early in this decade--moving from a field of many small firms, to the set of the Big Three. Nowadays, it seems that launching a new spreadsheet or word-processor requires more resources than GM or Ford would need to launch a new minivan. Producing a multimedia title, by comparison, is a shoestring operation.
But unlike 1900, when you had a good 30 years to bootstrap an auto company, or 1980, when you could dawdle for several years in a PC-software startup, in the '90s, the window of opportunity shuts at a much higher velocity. The thundering herd of financiers, Hollywood types, and giant-media conglomerates is pouring in capital to raise the barrier to entry. There are several ways they are seeking to gain a competitive edge. One is to follow "high production values" (slick film-industry standards for special effects, video, lighting, acting, and screenplay) and make the competition look amateurish by comparison. Professional screenwriters and actors are increasingly part of the development team. Another way to gain an edge is using the name recognition from a bankable star or famous person, even non-glitterati like William Colby. Finally, there is always the edge provided by proprietary technology.
The premier example of the venture-funded approach is the startup called Rocket Science, the sardonic, yet aptly named Silicon Valley company composed of some high-octane talent, including Peter Barrett (inventor of the industry-standard CinePak video-compression scheme), Bruce Leak (one of the developers of Apple's QuickTime digital-video architecture), Michael Backes (graphics supervisor for the movie Jurassic Park), Ron Cobb (creative director for the movie Alien) and Rich Cohen (creator of visual effects for the movie Terminator 2). Venture-capital funded at a $7 million valuation, Rocket Science plans to debut its new generation of gaming technology for the coming holiday season. According to the newsletter Digital Media, the production budget for the first three titles averages around $800,000 each. This is about eight times the starving-student budget used by The Journeyman Project.
Less visible is a small startup called Open Mind, cofounded by Mitsu Hadeishi, a Harvard physics graduate who did a stint as a game programmer at Electronic Arts. Originally based in LA but now in southern Oregon, Open Mind is working on a development project for well-known game company Trilobyte, publisher of the top-selling interactive multimedia title, The 7th Guest (which comes on two CD-ROM discs). Mitsu and his partner Doug Cutrell (a mathematics Caltech graduate) have been working for four years on an object-oriented, knowledge-based environment for designing more human-like interactions within the multimedia experience. Mitsu is tight-lipped about the project, but acknowledges that he is an expert in neural-net algorithms and that development is hosted on Windows NT.
As with the original California 49ers, participants in the new gold rush are finding that things don't always pan out. Many multimedia titles cannot stand on their own; instead, up to 80 percent of titles are sold bundled with hardware (CD-ROM drives or upgrade kits), resulting in revenues far below the standard list price. This is partly due to the phenomenon of "shovelware," in which seemingly random piles of existing content are shoveled onto a CD-ROM, without regard to consumer benefit. Some hapless consumers are getting hit on the hardware front as well. Walter Mossberg of the Wall Street Journal reports that anywhere from 25 to 40 percent of PC multimedia upgrade kits are returned to the store by frustrated consumers unable to get the IRQs and INTs straight. And those developers seduced by Microsoft's impressively large numbers for the Windows platform are finding that the cost of supporting Windows titles compared to the Mac ranges from 10:1 to 300:1. For some developers as many as one-third of PC support calls fall into the category called "Packard Bell problems," due to glitches encountered on these cut-rate but apparently not-quite-compatible machines. Also, in eyeing the attractively large installed base of CD-ROM drives, it is unclear how many are of the older single-speed variety, and therefore less likely to keep up with the increasingly stringent demands of digitized video and sound.
Lastly, in case you were tempted to stockpile blank CDs for future manufacturing, the consensus of many observers is that CD-ROM is a transitional (and barely adequate) medium for delivering multimedia content. According to the pundits, the future delivery vehicle is the Information Superhighway (or InfoBahn, or whatever it's being called this month). Audrey McLean, founder of KidSoft (distributor of multimedia titles for children), calls CD-ROM the "SCSI Highway," and considers it an interim substitute for the real thing.
Unlike some multimedia authors who've packed their bags and headed for California to be closer to the action (as did the author of the popular "HellCab" title, who moved here from New York City), you may be tempted to sit this one out. If so, what are the implications of this gold rush for those on the sidelines?
One implication is the influence on PC-system architectures. There is a parallel with the effect of desktop publishing (DTP) applications on PC-system designs over the last ten years. Although DTP was only used by a small segment of the overall PC market at that time, the hardware demands of this application area drove technology development for almost a decade. Aspects of today's mainstream PCs, such as VGA and SuperVGA-resolution graphics, square-pixel displays, outline fonts rendered on the fly, device-independent printing, virtual-memory operating systems, and so on, are in large measure due to DTP's stringent demands. Even certain Windows API functions, like the GDI function ExtTextOut(), were added at the specific request of DTP implementors (in this case, the authors of Aldus PageMaker). These enhancements were added even though the majority of PC users at the time were quite content with the more modest requirements of Lotus 1-2-3 running in character mode.
Likewise, although many users are now quite content with today's mainstream desktop configuration, you can brace yourself for technological evolution along lines dictated by multimedia needs. One basic requirement is incredibly high throughput: Without compression, a 640x480 24-bit image, displayed at 30 frames per second, means an uninterrupted data rate of 27 Mbytes per second. Satisfying this requirement will impact all aspects of PC system design. First, of course, is hardware support for compression and decompression. It's possible that MPEG decoders will be added to motherboard designs over the coming years. Graphics adapters may have hardware support for RGB-to-CUY conversion (the CUY schema is used by video input devices and is more amenable to compression algorithms). And, just when you've made the decision between VESA local bus and PCI, you'll have to choose between VESA Media Channel (VMC), a 132 Mbytes-per-second bus architecture that supports digital motion-video processors, and Intel's competing Shared Frame Buffer Interconnect (SFBI). In the area of drive interfaces, IDE and SCSI no longer make the grade; you'll need, at minimum, the 13 Mbytes/sec data rate provided by SCSI-II with the "fast" and "wide" extensions. Operating-system file structures and algorithms may be modified to support uninterrupted large-data streams. The thirst for large amounts of RAM will make NT's current demands seem paltry. In networking protocols, neither TCP/IP nor IPX/SPX will hold a candle to ATM. So a few years from now, when you are trading in your doddering 90-MHz Pentium/PCI machine for one of these high-bandwidth multimedia models, you can blame the Gold Rush of '94.