Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!hao!ames!aurora!labrea!glacier!jbn
From: jbn@glacier.STANFORD.EDU (John B. Nagle)
Newsgroups: comp.graphics
Subject: Re: Data Compression
Message-ID: <17224@glacier.STANFORD.EDU>
Date: Sun, 22-Nov-87 01:50:52 EST
Article-I.D.: glacier.17224
Posted: Sun Nov 22 01:50:52 1987
Date-Received: Tue, 24-Nov-87 01:37:20 EST
References: <619@applix.UUCP> <305@etn-rad.UUCP> <3387@adobe.COM> <308@etn-rad.UUCP>
Reply-To: jbn@glacier.UUCP (John B. Nagle)
Organization: Stanford University
Lines: 28
Keywords: video compression


     The basis of video compression is the notion that each frame is usually
a lot like the last one.  There are various schemes for sending the changes.
I've seen images transmitted using commercial gear which compresses a video
signal to fit into a 56KB line.  The system really works only for mostly
static images; whenever anything moves, the area of change is shown as
large squares of uniform density and color.  As soon as things begin to
stabilize, the squares are subdivided into smaller and smaller squares
until each uniform area occupies only a single pixel.  This process
takes about a second after a major change in the image, less for
smaller changes.  A person talking without moving their body too much
looks almost normal.  At the other extreme, panning the camera produces
complete breakup.

     Considerable work is underway on finding better ways to compress
video.  One motivation behind this is the desire for a better approach
to high-definition television.  The proposed scheme from Japan requires
three times as much bandwidth as present television signals, and this is
unacceptable for broadcast.  As yet, RAM still isn't cheap enough to 
allow a frame buffer in every TV set, (although some high end units have
one now), but it is recognized that that day is not far away, and 
work on compression goes on at the MIT Media Lab and at consumer electronics
companies.

					John Nagle