1992 Discover Awards: Computer Hardware and Electronics

Thursday, October 01, 1992
It’s kind of like taking water out of orange juice to make concentrate, and then adding water to turn it back into orange juice later on. That’s how Todd Townsend of Compression Labs describes the process of video compression, the technology that makes AT&T;’s new VideoPhone 2500 possible. The VideoPhone is the first practical way to transmit audio and full-motion color video over existing phone lines.

A videophone has been the stuff of science-fiction fantasy for more than a century; shows from The Jetsons and Flash Gordon to Star Trek had a videophone in a starring role. Although technologically feasible since 1964, when AT&T; unveiled its Picturephone at the New York World’s Fair, no one could figure out a way to make a practical model.

In the 1980s Sony and Mitsubishi introduced inexpensive plug-in- the-wall video telephones, but these could only send black-and-white still images, and conversation had to be suspended during picture transmission. Videoconferencing became a part of corporate culture in the 1980s but required specialized networks and expensive equipment.

The VideoPhone is an ordinary-looking business telephone that plugs into any standard phone jack. The only difference is the flip-up video screen, which displays a live picture of the person at the other end of the call, and a video button, which allows either party to block the use of the video screen at any time.

To configure their VideoPhone, AT&T;’s research arm developed a new high-speed modem that pushes information about sound and pictures across a phone line. The biggest obstacle to a workable videophone, however, was finding a way to cram data-hungry video images--which eat up 92 million bits per second (a bit being the smallest possible unit of information, the 0’s or 1’s of computer language)--through the existing national telephone network. This network can only handle 19 thousand bits per second.

To do this AT&T; turned to Compression Labs in San Jose, California. Townsend, who holds seven patents in digital telephone transmission, was the architect of the VideoPhone’s compression technology. The first and easiest step he took was to reduce the number of images captured by the system’s built-in camera from 30 per second (the standard for broadcast television) to 10. That eliminated about 60 million bits right there. Even at 10 frames per second, the phone’s fixed-focus camera produces an image that’s equivalent to the output of most camcorders. Not bad, considering the camera’s lens is only about the diameter of a penny.

Next Townsend and his team shaved off another chunk of information by using a small, 3.3-inch-square TV screen on the VideoPhone. The smaller the screen, the fewer the pixels needed to create an image. While conventional TVs have a resolution of 300 by 400 pixels, the VideoPhone screen has a resolution of 112 by 128 pixels.

But the most significant innovation in video compression was Townsend’s codec, a coding and decoding device built into each VideoPhone that ignores everything but the most crucial parts of a moving image. The codec processes only the parts of the image that change from instant to instant, enabling the VideoPhone to transmit and receive only the barest minimum of information.

The codec divides the entire screen into 224 blocks (each is 8 by 8 pixels) and analyzes each block ten times a second. If the image in a block hasn’t changed from one frame to the next, the codec doesn’t transmit the data for that block repeatedly. Instead, the codec keeps the same image on the screen until it changes. For example, if there’s wallpaper behind a caller, no new information needs to be sent about that background, unless, of course, the caller’s head covers it, or it catches fire. If there’s a slight change in the image from one instant to the next--lips moving, for instance--the codec knows to send just the difference between the two images.

If there’s a major change in a block--like a hand waving--the codec recodes that block entirely and the new image completely replaces the old one. But as most videophone calls display a person’s face, the majority of the VideoPhone image stays the same during a call.

In the end, the video portion of the signal is compressed from 92 million to just 11,200 bits per second. More than 99 percent of the original picture data is removed, a compression factor of 6,000 to 1.

The VideoPhone transforms the compressed data into an audible sound that can be sent over standard phone lines--for the same price as a conventional phone call. The sound is not unlike the screech heard when you accidentally call a fax machine. It’s an absolute forest of tones and notes, very complex and interwoven, says Townsend.

At the other end of the phone call, the codec in the receiving VideoPhone listens to the screech and decompresses the signal--adding water back to the orange juice. The resulting image isn’t up to the quality of standard television, but it’s surprisingly clear.

Townsend’s work came to a climax last spring, when he placed a VideoPhone call from San Jose to the AT&T; lab in Indianapolis. When the connection came up, recalls Townsend, I could finally see the guys at the other end who I’d been working with for a year and a half. Everyone crowded around the camera, whooping and hollering.

The video screen is playing an ever-increasing role in the way we work, play, receive news, form opinions, and, now, in how we will communicate. Certainly there’s much to be gained by seeing the person at the other end of the telephone line. Grandparents will be able to see the new baby minutes after birth, and scattered families will be able to share holidays together. Video telephones may turn out to be a colossal gimmick, like quadraphonic stereo, but then again they may send the voice- only telephone packing in the same way push buttons punched out the rotary dial. If the VideoPhone is to become truly popular, however, the $1,500 retail price tag will have to drop dramatically.

Jon Krakower and Jon Sedmak, product designers at Apple Computer in Cupertino, California, for the development of the Macintosh PowerBook line of portable notebook computers. These sleekly designed computers feature all the power of a desktop Mac in a lightweight, affordable, easy- to-use package. Each PowerBook sports an upside-down mouse, or trackball, that enables users to maneuver quickly around the screen. The designers made user comfort and convenience a priority, a bold achievement in the world of portable computing.

Jerry Erickson, research and development section manager at Hewlett Packard in Corvallis, Oregon, for the development of the HP 95LX Palmtop PC. Weighing an amazing 11 ounces, this calculator-size computer has all the processing power of a conventional desktop model. It includes built-in Lotus 1-2-3 spreadsheet software, a time management system, memo writing capability, and an easy-to-use phone book. The HP 95LX affords the user on-the-go information and analysis anywhere, anytime in the palm of the hand.

Kamran Elahian, former chairman of Momenta in Mountain View, California, for the development of the Momenta Pen-top portable notebook computer. This unique modular computer runs all DOS and Windows applications and lets owners use either a pen or a traditional keyboard as an input device. In addition to opening the world of computers to the typing-illiterate, the pen-based input system allows people to use the Momenta in meetings and other settings where a keyboard is obtrusive. The Momenta captures electronic ink to take notes, draw graphics, and transcribe handwriting into text. While only the size of a three-ring binder, the powerful Momenta even includes a fax machine.

Yoshitaka Ukita, product designer for Sony in Tokyo, for the development of the Data Discman electronic book player, the first product to unleash the power and versatility of CD-ROM disks and electronic books in an easily portable package. The Data Discman comes with a full lineup of software titles covering a variety of subjects, including education, entertainment, business, and health. Each disk can store more than 100,000 pages of text, 32,000 graphics, or up to 5.6 hours of sound. No longer is retrieval of the wealth of information now stored on CD-ROM limited to the PC-literate and their desktop-anchored peripherals.
Comment on this article