Your site may soon look naked without a few good 'toons

Ever since the 1950s, when television rapidly exploded to nearly total penetration of the population, vast sums of capital have been spent hunting for the next big video thing: high-density TV, cable, interactive TV, videophones, even 3-D TV.

The latest incarnation of this passion has been the parade of projects promising to bring video to the Internet. Almost every month brings a new product claiming to support some business-oriented video application or another: marketing (landscape views from hotels), public relations ("talking head" concept presentations by CEOs), training (on-line videos), sales (videoconferencing sessions with sales personnel), and so forth.

The barrier against which all these products must prove themselves is the requirement for far more bandwidth--that is, connection speed--than is available on today's Internet. All the products mentioned so far have to squeeze their feeds through 28 Kb modems--the computational equivalent of a needle's eye. So far, no one has found a method without disagreeable trade-offs, such as tiny screen sizes or jerky motion or a rigidly static camera. These approaches usually end up consuming lots of bandwidth anyway, pushing out other applications.

Although it is possible that the general public will put up with images of such poor quality, a prudent wager on Internet video necessarily entails a side bet that much faster connections are going to be delivered to end users at an affordable price. This might happen, thanks to cable and satellite modems--then again, it might not. Both the telcos and cable companies are fighting off lethal threats to their core products while struggling to reform sluggish, monopolistic, consumer-hostile, institutional cultures. The satellite-wireless and cellular-wireless industries are starting from zero in terms of built structure, and their costs are high.

Obviously, it is not wise to spend money on services that depend on the delivery of major increases in end-user bandwidth to large segments of the population within the next five years.

How then, without video, to produce Web sites that live up to the promise of all those dynamic applications? There is a way. During the five or so years that it takes the industry to make Internet video practical, the whole moving-picture market might be delivered into the hands of another medium altogether: animation.

In the off-line world, distribution and market constraints have kept animation confined to a tiny industry--an occasional feature from Disney, some Saturday morning cartoons, and a few dozen TV ads. On-line, the strengths of the medium make such a natural match with the needs of the Net that a grasp of animation might be essential to running a business with a Net component before the end of this year, let alone five.

The difference between video and animation is that unlike video, animated imagery can be made machine-readable. A video camera pointed at a bouncing ball "sees" nothing but arrays of dots with changing light intensities. Despite decades of work on pattern- and image-recognition, a computer still cannot look at a photo or video and name its contents (even if the content is defined as minimally as "edge" or "corner"). By contrast, the instructions in a computer animation of a bouncing ball will list the name of the file containing the ball prototype specifically, making it trivial to retrieve. In short, animations are more intelligent than video.

One implication of its higher I.Q. is that animation needs to consume very little bandwidth. Once a vocabulary of geometric archetypes or polygons has been downloaded (or even distributed on CD-ROM, as some on-line game companies do it), a server can deliver an animation of a bouncing ball simply by sending the name of the appropriate routine ("bouncing ball #16") straight to the client. When the client gets that command, it just pulls the ball #16 archetype off its disk or out of memory, calculates how this case should look, and puts that picture on the screen. The command to pull the ball routine shouldn't take more than one second to download at current speeds; a short, fuzzy video of a bouncing ball could easily take several minutes.

Each succeeding frame of an animated image can be recalculated locally at the user's PC, at whatever speed is needed to generate a quality animation display, and without necessarily requiring so much as another byte from the originating server. Given enough processing power and memory, you can have full-screen high-quality animation feeds in hundreds of colors and with dozens of tumbling objects, perceived through a constantly changing point of view. And with plenty of bandwidth left over for telephony and supplementary data feeds, it can fit easily over a 28.8 Kb connection--although that depends on such things as the quality of the phone lines and which applications are running in the background.

An even more important property of machine-readability is that it allows animated objects to interact with other animated objects and with humans. Consumers surfing to the site for MYski Inc., of Los Altos, Calif., can design their own skis (colors, logos, text) and then rotate the finished design to see how they look. "Without animation, it would be difficult for our customers to see what they were buying," says company president Chris Jorgensen. "It's the closest we can get to hands-on through the Internet channel." Users at can design, view, and then mail animated greeting cards. AniMagicians, an animation group in Boston, has the capability to build large Web stores that will support 3-D customer walk-throughs. Maintenance engineers at Duke Power, in Charlotte, N.C., can call up 3-D designs on the company's intranet, combine them into assemblies, and then rotate the assemblies, inspecting them from any desired point of view.

One of the most promising applications of animation is the reuse of the same set of animated objects (images) throughout a production cycle. Given machine-readability, once the basic designs are completed, anyone at any stage of the product life span--manufacturing, packaging, transportation, vendor displays, customer assembly, customer support, maintenance, moving, and disposal--can run those images in sequences and perspectives that make sense in their current context.

For example, suppose the originating designer of a dollhouse kit draws two walls that must meet at a joint. Those images, as well as 3-D files, might then go to the assembly process designer, who would use them to locate and size the tabs and slots needed to make the connection. They might then help manufacturing organize all the tabs on the same sheet of paper, vendors design the most appealing product displays, and customers see what the product looks like before assembly (from any angle they like).

After purchase, customers might assemble the kit by following an animated assembly sequence on the vendor's Web site. If they still had problems, customer support would use the same images to walk them through the process. (MYski already reuses in a small way: When a customer has designed a pair of skis on the MYski Web site, the specifications go from the site right to the factory floor.)

Animation can also be used to show--in a clear and graphic way that could never be duplicated with static images--how ordinary data changes with time. For instance, Toyota Motor Sales, U.S.A., in Torrance, Calif., contracted Digital Evolution to build a corporate intranet that uses animation technology such as Java. The intranet's graphic interface allows managers to perform tasks such as monitoring inventory at a glance.

An even more ambitious example is an information display system announced last October by Sandia National Laboratories in Albuquerque. The system organizes all published science pieces into a single animated landscape. The program examines a list of publications and creates a virtual landscape in which articles that are closely related to one another are placed more closely together in the display. The display is then animated to show how the relationship between articles changes over time. Over the course of a year, you could build up an animated landscape that shows trends in research as rising and falling waves; a scientist might follow the progress of an experimental technique by looking at the peaks and valleys in the landscape. Instruments such as these might one day be essential to navigating the information highway.

Finally, animation is potentially more flexible than video because every aspect of the medium can be controlled. "You don't see video footage even in arcade games" (where bandwidth is not an issue), points out Aaron Shi of AniMagicians. "People see video every day. The look is too common. What they want are fantasy environments." It is particularly useful to employ moving pictures rather than words at Web sites whose visitors might have a weak command of English.

The constraints on animation are fast processors, large memories, universal standards, and creation software--obstacles the industry is attempting to overcome. In 1996, chip companies shipped over 8 million 3-D graphics accelerator chips. In January, Intel announced a microprocessor designed to work with multimedia applications. By the end of the decade, every new computer sold will come with some level of 3-D rendering capability that at the beginning of 1996 was available only on "visualization" workstations and supercomputers.

In 1995, a committee of 60 companies, including Sun Microsystems, Netscape Communications, and Silicon Graphics, came together and agreed on a standard set of programming terms (VRML 2.0) for recognizing and executing animation, which make it possible for everyone to write, send, and receive animation without first buying into some proprietary scheme. As of this writing, Netscape Communications is planning a second-quarter release of VRML-capable browsers, and Microsoft will follow soon after. Finally, the Java applet system being promoted throughout the Web is ideal for transmitting animated "objects" to desktops.

The one constraint in animation that as yet has no quick fix is production cost. Programs that allow lay persons to design and manage animated sequences are only now beginning to appear. Silicon Graphics Cosmos suite (Mountain View, Calif., $2,300 for Silicon Graphics version, but a Windows NT version was just released, pricing to be determined, 800-800-7441,, which converts 3-D CAD designs into VRML for distribution over intranets and the Internet, is one professional entry. The program WebPainter, by Totally Hip Software (Vancouver, B.C. $99.95, 604-685-6525,, is an entry-level tool for exploring the options.

Sooner or later, someone is going to figure out how to generate imagery that a computer understands at least well enough to figure out the faces and edges directly from a videocam. Once that happens, the wall between video and animation will collapse; video will require no more bandwidth than animation, and animation will have the cheap production costs of video. All the applications that can be served by either medium will combine into the same product stream.

That should make for some pretty exciting Web sites.

Animation Versus Video

Bandwidth requirements:
Video: Video requires at least 10 times the current speed for reasonably fast downloading of video clips.

Animation: Once a "library" of images is stored on a PC, complex animations can be conjured up in seconds via a standard connection.

Video: A video clip is fixed; there is no good way for a user to interact with images nor for images to respond to other programs.

Animation: Animation can be controlled by the user video-game style, or it can be made to interact with other programs and data.

Video: Video can't be adapted to illustrate different points and serve different functions.

Animation: A single animation can be easily modified to fit different applications, such as engineering design and customer training.

Fred Hapgood is a freelance writer based in Boston.