Home
Next Event
Past Events
Resources
Volunteers
What's New
ACM SIGGRAPH
Digital Video: Present and Future

Robin Wilson
DiviCom

November 20th, 1997

Notes by John Peterson

Robin spoke on current and future technologies in digital video. One of the principle issues in digital video is compression, because of the tremendous bandwidth required for digital video images. Some typical bandwidth measures that indicate the problem:

 Today's NTSC 115 MB/s
525 Component video  270 MB/s
HDTV   1 GB/s

A typical 6 megahertz video channel has only 20 MB/s of bandwidth space available.

Video compression has been evolving for the past 60 years:

1930's Interlacing   2:1
1950's NTSC Color   5:1
Today Digital compression 100:1

Interlacing, one of the oldest techniques, sends half the motion half the time by alternating the scanlines drawn. However, it makes subsequent modern compression techniques hard to apply. NTSC shares bandwidth between color and brightness information, taking advantage of the fact that the human visual system is much less critical of color information than tonality. Thus, the color information is highly compressed and transmitted separately.

Many modern techniques take advantage of "lossy" methods that throw information away that the human visual system is not likely to notice, allowing much higher compression. Further savings are achieved by sending descriptions of the images rather than the images themselves, such as when a still image appears on the screen for a few seconds.

Most of the methods for compressing a frame of video are based on block DCT (discrete cosine transform) encoding. The DCT method converts a small block of the image from a spatial image to a frequency-based encoding. This data is re-arranged so the higher frequency details are removed without producing noticeable artifacts. The remainder is compressed using standard digital techniques.

The most popular digital video compression standard is MPEG (Motion Picture Experts Group). This standard has gone through revisions over time. The first, MPEG 1, was not specified in enough detail to allow for widespread broadcast and recording use. MPEG 2 was better defined, with both a system and a transport layer. Improvements that were to become MPEG 3 were later subsumed into the current MPEG 2 standard. Work on MPEG 4 is still underway, but Wilson described it as more of a "research project", still waiting for a breakthrough to achieve its goals. The current MPEG 2 standard supports a number of different formats and techniques that broadcasters and receivers must support. For example, it is possible for broadcasters to switch the image resolution on the fly. The MPEG 2 format is well accepted, and used in a number of applications such as direct broadcast home satellite dishes.

Because of the wide number of techniques and formats available in MPEG encoding, content creators can use a number of tricks to maximize bandwidth. For some applications, such as creating a DVD title, the compression can be done "by hand" to select the best method for a particular scene. Some types of content remain difficult to compress though, such as football and basketball games.

Goals for future digital video technologies include lowering the data rate even further with advanced compression techniques, transmitting video over the Internet, and directly manipulating and editing compressed video.


Back to Silicon Valley SIGGRAPH home page.

Send comments to the webmaster