DIGITAL VIDEO IS STILL AL DENTE
It’s better than it was, but it’s not done yet

John North is vice president of strategic products at ICTV/Inteletext Systems Inc., an interactive television company in Santa Clara, CA.

To hear the workstation marketeers, their software vendors, and especially the trade press talk, digital video and multimedia are a fait accompli, right on schedule to overrun the tools and techniques of a 100-year-old film industry and a 60-year-old television industry.

The truth is not completely divorced from this vision. From a certain perspective, there’s no question: The Mac “does” video. The PC “does” video. Silicon Graphics machines “do” video. Next “does” video.

But the vision is incomplete in some significant ways. What I hope to do here is simply identify what’s been omitted from a real “workstation-does-video,” with the hope that such a system can emerge as more whole, complete and useful to the business of film and video. The purpose is to match creative and economic forces with the existing technical standards for a deliverable product.

CART BEFORE HORSE

Technical interface standards and methods of operation for telefilm and televideo are long established, particularly with regard to machine control, signal envelopes, reference timing, chromaticity values, media indices and edit decision lists.

But unfortunately, the counselors of computerdom have chosen to address the multimedia technology issues not by using a computer as controller, but rather by using the computer for image storage and processing. I submit that this is putting the cart before the horse.

Their decision to import/massage/export video frames could be considered bad data management: a single video frame can consume on the order of 4 megabytes, which is not bad until the user is compelled to push 30 frames through the machine every second. A more sensible solution is to use the computer as a controller and take advantage of the considerably smaller files that are generated as a product of connecting the computer to the outside world — i.e., to the mature systems already in place for producing film and television.

THE MONEY LINE

The creation of sustained revenues —the money line — that connects the “system” on one end to anybody who has to produce, edit or syndicate/broadcast film or television products for a living, on the other, is uniform compliance. The still-unrealized power of the video workstation for the average Joe will come with simplification of the user interface and automation of critical technical standards. Once completed, a full-featured video workstation will give a single unified means of accessing (ergo controlling) a larger field of equipment and materials.

The console for a high-end video editing facility can easily stretch 12 to 15 feet. On this console are remote control panels for production switchers, edit controllers, one or more special-effects devices, an audio console, a character generator and utility geegaws like waveform monitors and vectorscopes, title cameras . . . whatever.

In a large facility, it’s a bunch of stuff hanging from remote umbilicals to a central machine room that houses the incredibly expensive, real-time, multi-layer compositing equipment, aka “heavy iron.” Is it so hard to imagine a single, icon-based computer workstation controlling all these devices?

Here are the incomplete or otherwise missing factors to making such a system a reality.

CONTROL AND SYNCHRONIZATION

There are two key barriers to realizing the improved computer video workstation. Repeatable event synchronization between NuBus, SCSI, serial, main memory and co-processor operations in the Macintosh is now virtually impossible. And there are not enough pieces of video equipment that have open, nonproprietary serial communications.

The problems are not solely on the workstation side of the equation. Save for video tape machines, video manufacturers usually only include full RS-422 serial-port connections for external control on the most expensive of the production switchers and effects-generating equipment. Time base correctors, utility routers, smaller mix/effects switchers . . . every piece of equipment shipped from a video vendor needs a serial port, and vendors are afraid to open up their equipment to serial connections.

Why? For one thing, they fear the centralized control that serial connections enable — they require extra engineering and their margins are already thin. They also (wrongly) believe that by keeping their systems closed to central control, they will be able to protect high-end systems from competition.

THE SIGNAL ENVELOPE

A signal envelope is the set of analog waveform characteristics that define a specific video format (NTSC, PAL, SECAM). This envelope is an amalgam of edge shapes, rise times and horizontal/vertical blanking intervals. Digital video is an occasional, increasingly common and principally high-end production format. But that which is digital will not become analog for distribution and broadcast until some unknown date when HDTV broadcasts begin.

The signal envelope is only as good as the equipment you pass it through. In an online edit session, signal quality is diligently reviewed on a scene-by-scene basis. Workstation control of the video system must include remotely accessing the adjustments for these external signal conditioners.

Reference timing. Signal envelopes are referenced against time. Without going into great detail, values such as genlock, sub-carrier burst-lock and SC/H phase coherency are important time values that affect the conversion of a high-resolution, noninterlaced computer picture into NTSC or any of its derivates. Workstations to date have virtually no recognition of these signal states, much less the capability to alter them.

Chromaticity values. There are many more colors in the visible spectrum than can be reproduced in video. In fact, the range of colors varies considerably on the finished media: light, print film, movie film or television. Even within the range of film stocks are wide variations on chroma potential.

These limits are defined both by the physical capabilities of the system and in law by the Federal Communications Commission, giving rise to the term “legal video.” Legal video encompasses the factors described both here in the chromaticity values and those in reference timing, above.

Software filters included with Adobe Photoshop and the Video F/X product save the day with look-up tables for legal NTSC colors. Kodak’s PhotoCD technology, though not well integrated into any video systems so far, is a great standard-bearer for outputting a single file or file series into multiple media. The advantage of PhotoCD is that color palettes can be prebuilt specifically to the chosen recording or presentation media.

What’s needed to solve the “workstation-does-video” paradox is the capability to reference legal chroma values against an external, systemic, color burst reference inside the computer.

MEDIA INDICES

SMPTE (Society of Motion Picture and Television Engineers) time code for video and frame edge numbers for film are the means by which individual, physical frames may be logged electronically against scene, take and edit decision list (EDL) data. After many years, SMPTE code fields are now increasingly present in media sequencing software.

One notable omission is Adobe’s Premiere (though SMPTE time code has been promised in a future release), but the grand omission of the era is most certainly from Macromedia né Macromind, which has told me flat-out on several occasions that Director will never itself be SMPTE-aware — though it is possible to juxtapose SMPTE data accessed and decoded independently of Director through various external commands.

Details, details. Film edge numbers are converted to and from SMPTE, as are MIDI time stamps (which contain both content and control data). There are four information blocks in each SMPTE byte equivalent that vertical blanking synchronize, hours, minutes, seconds and frame bits, space for user notations, and cyclical color field IDs. (Color fields must be cut in sequence in order to avoid a chroma glitch in the finished product.)

Edit decision lists. EDLs, simply the line- and tab-delimited ASCII representation of SMPTE head and tail frame data, are tightly tied to SMPTE in significance. The EDL allows the preservation of a user’s editorial decisions across platforms; i.e., between workstations and suitably qualified professional video edit systems that often interface with the aforementioned “heavy iron.”

To some, SMPTE and EDLs are simply details. But since all these things are referenced against each other and are totally interdependent, the bottom line is “do it all, or why bother?”

INTERACTIVE PROGRAMMING COMETH

Multimedia workstations will be fundamental to the creation of the juxtaposed real-time, broadcast-quality video materials with interactive data. Completed interactive systems will enable home viewers to access, directly through their televisions or other devices, a directory of multiple choices for response.

An ideal interactive system will incorporate more distribution input and output, video sources, image libraries and networked data in endless permutations than has ever been handled to date.

Adding to the suite. In order that such broadband interactivity is possible, there is a critical need for a broad systemic overview and control from workstations located in interactive operating facilities (what will become the interactive equivalent of a cable head end), servicing multiple personal clients simultaneously.

To bring the computer to a level of video signal and machine control literacy, machine control, event synchronization, signal envelope integrity, SMPTE time code and EDL look-up tables are all absolutely necessary components of video workstation operating system and applications software.

It is only when workstation video products are able to recognize and address (internally or peripherally) these long-established functions that the consumer digital video market can explode. The vision, the promise for personal digital television is access; access through the network — including long distance carriers, local phone companies, cable TV and the Internet — to retrieve and operate upon the oceans of sound, picture and textual information that are and will be available.

Mass-scale markets cannot or will not happen until all the factors of remote system control happen from a single graphical user interface in the professional community. Once the control paradigms are established, it is a relatively easy task to simplify the user interface by further abstraction.

MISDIRECTING COMPUTER POWER

Since specialized video equipment is externally controllable and of extremely high quality, I believe that the power of the computer has been disproportionately directed away from an eminently desirable task: simplifying the control of a very complex system.

The full integration of system control and image processing feels inevitable. The fact that it hasn’t occurred to date is much more the matter of short-term market forces and eager technologists making “cool tools,” postponing the hard work that will allow their customers’ workstations to interconnect with video equipment long installed around the planet.

The film and video industries, with only nominal contributions from desktop computers, generate many billions of dollars in revenue each year. The expansion of this revenue base is dependent on simplified control that puts system literacy into the electronic domain, and allows the creative mind the means to produce finished, distributable products. The workstation video technology being touted today simply isn’t good enough to complete in the job it starts.

John North