The Cole Papers, Audio-Video, April 1997

Real-time video and audio gaining rapid web acceptance

Hollywood is taking over the Web.

A production worthy of anything ever dreamed up in Tinseltown awaited about 650 people drawn to the Progressive Networks RealMedia Conference March 3-4 at the Hyatt Regency San Francisco Airport. Powerful rock music and colored searchlights greeted a crowd that represented half the states and 16 countries.

It was an appropriate scene for a conference that focused on the use of audio and video technologies to bring the Web to life.

"Working together, we'll turn the Internet into the next great mass medium," said Rob Glaser, chairman and chief executive officer of Progressive Networks. Glaser is hoping his company can lead the way into that next great mass medium with its RealMedia product.

Progressive is the company that gave web developers RealAudio, the streaming technology that made sound a feasible part of web design. Rather than having to download a large sound file and then play it, RealAudio users can listen to an audio stream as it is downloading, avoiding the long wait Internet users previously experienced when they wanted to hear sound files.

RealMedia is Progressive Networks' next step: streaming video. RealMedia incorporates RealAudio with Progressive Networks' new product, RealVideo. With RealVideo, rather than downloading a large QuickTime or AVI video file, users can watch pictures as they are downloading, making web video a more television-like experience.

Making the Web more like television seemed to fly in the face of figures cited by Jim Banister, vice president and general manager of Warner Bros. Online.

Studies show that 58 percent of people who go on-line watch less television, he said, with market research showing that these viewers-turned-surfers may be watching as much as 15 percent less television than formerly.

It apparently doesn't seem so odd to Hollywood filmmaker Spike Lee, who created three short films for the RealMedia site when the video product debuted in February. "We've had over 100,000 people watch these short films in a three-week period," Glaser said.

Glaser likened the development of the Web to the evolution of radio and television.

"Radio is the grandfather of mass media," he said. In the 1920s, commercial radio was in its infancy, born when Kdka in Pittsburgh offered live updates during the 1920 presidential election.

In the 1950s, television entered adolescence, with such epochal events as Elvis Presley's appearance on The Ed Sullivan Show.

In September 1995, Internet broadcasting was in its infancy with RealAudio's broadcast of the first live baseball game on the Web. After playing a clip of that broadcast for the audience, Glaser observed, "It's interesting that after a year and a half, that feels like a historical document."

Two years ago, there was no RealAudio, Glaser said; today, there are 15 million users. "Where will video be in two years?" He forecasts rapid growth, predicting that "1997 will be to video on the Internet what 1995 was to audio."

Bruce Jacobsen, president and chief operating officer of Progressive Networks, said in addition to the 15 million RealAudio players already in use, about two million are being downloaded from the company's site every month.

Video production tips
Much of the conference focused on speakers playing audio and video clips from their web sites and discussing how they were produced. Other speakers gave tips on audio and video production.

Norm Meyrowitz, executive vice president of Macromedia, the company that gave birth to the popular Shockwave plug-in for web browsers, suggested animation as a logical choice for web video because animation is more natural for the low frame-rate typically delivered on relatively slow modem connections.

Animated shows such as Beavis and Butt-Head or The Simpsons could be delivered over a $99 modem, Meyrowitz said.

Philip Rosedale, general manager of the applications group at Progressive Networks, said animation is ideal for the RealVideo encoder because it lends itself to the "slide show" format of one frame per second or slower, which allows the encoder to put more bandwidth into the associated sound.

"It's the audio that drives the experience," Rosedale said.

Rosedale gave other tips for video production:

Use a tripod and a stabilizing camera. It might seem obvious to some, but failing to do so is a common mistake among novice videographers, he said.

Use a medium angle. Too wide an angle allows background movement to invade the image. That unnecessary action causes the encoder to produce "artifacts," which show up as blocks of color or shadow images.

Too tight an angle can make the subject move too much in the frame. A medium angle allows for much of the frame to remain still while having some motion to bring the subject to life.

Don't do dynamic zooms. Though that effect works dramatically on television, the constantly changing image confounds the compression technology of the encoder.

He also suggested avoiding transition effects such as fades and dissolves. Straight cuts from one scene to another are easier for video encoders to handle without adding undesirable video effects.

The RealVideo encoder (and other video encoders) depend on comparisons between adjacent frames to achieve file-size compression. Any unnecessary movement decreases the encoder's ability to reduce the size of the file, and larger files are bad news in today's web environment -- bandwidth still rules all web activity.

People being videotaped or filmed for a web broadcast should be more theatrical in their movements, Rosedale said, because bandwidth limitations require video images to remain relatively small.

Rosedale likened it to watching television from across a large room: Small details are not noticed; dramatic movements get the viewer's attention.

The primary goal when encoding either a "talking head" video or a music video is to get the video smooth enough to look synchronized with the audio, Rosedale said. Though the perception is purely a personal matter, he suggested 7.5 frames per second would be sufficient to achieve lip sync.

Steve Mack, senior audio engineer for Progressive Networks, offered tips on dealing with audio content. Oddly enough, he said, when choosing between a mid-range and a professional-quality sound card, the mid-range might be a better option.

The professional cards, he said, typically sample sound at a rate of 11 kilohertz. RealAudio and the audio portion of RealVideo require a sampling rate of 8 kilohertz. The professional-quality sound cards are capable of squeezing their samples down to 8 kilohertz, Mack said, but they don't do it very well.

Web sound engineers might just as well spend their money on a mid-range sound card and get 8 kilohertz from the beginning, he said.

Compression and equalization are key factors in producing good audio content, Mack said. Compression entails reducing the dynamic contrast between the loudest parts of the audio and the quietest audible parts. Equalization involves enhancing selected frequency ranges and decreasing others, just as one would with a home stereo system.

Both can be done with either hardware or software, but Mack prefers hardware. For one thing, he is an audio engineer, and he likes to have slides to slide and knobs to turn. For another, software cannot do the processing with a live audio stream; only hardware can help with a live audio broadcast.

Josh Berman, webmaster for Kcrw radio in Santa Monica, Calif., advocates manipulating the audio file before encoding it. "The more pre-processing you can do going in, the better," he said.

Mack stressed the importance of investing in good equipment.

"If you buy cheap equipment, it's going to sound bad," he said. "Don't come crying to me with your problems. Spend a little money."

-- David Galloway

Macromedia Inc.,
(415) 252-2000;
Progressive Networks Inc.,
(206) 447-0567.

Real-time video and audio gaining rapid web acceptance

From THE COLE PAPERS, April 1997, Copyright © 1997, All Rights Reserved.