Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 11/08/85; site unccvax.unccvax.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!mcnc!unccvax!dsi
From: dsi@unccvax.UUCP (Dataspan Inc)
Newsgroups: net.analog
Subject: Re: Television over the telephone?  Is it possible?
Message-ID: <382@unccvax.unccvax.UUCP>
Date: Sat, 7-Dec-85 12:40:17 EST
Article-I.D.: unccvax.382
Posted: Sat Dec  7 12:40:17 1985
Date-Received: Mon, 9-Dec-85 03:50:41 EST
References: <1148@gitpyr.UUCP>
Organization: UNC-Charlotte
Lines: 56

> Does anyone out there know if it is possible to transmit a
> television signal over a standard telephone line?  It seems to
> .....

     It depends on what you mean by "television signal." If one means
live, NTSC-encoded composite television, forget it.  To meet all the
pertinent parts of FCC Section 73 with respect to broadcast quality
video, there is no way that a standard dial-up telephone circuit has
even 1/1000th the bandwidth required. (NTSC television requires a luminance
bandwidth of at least 4.2 mHz)

     However, there are many forms of compression which involve tradeoffs
of either time or spatial/contrast resolution that permit "television
pictures" to be transmitted via telephone lines.  

     In the temporal compression camp, we have single frame transmission
systems, exemplified by the kind my company makes (DataSpan) and several 
competitors, including Colorado Video.  The objective here is to transmit
just one still frame which, at the other end, would meet the FCC R&R for
a "single frame of television."  Our system requires ~130-180 seconds to
transmit a full frame of NTSC television, meeting and/or exceeding all
broadcasting standards.  The product is primarily used in the medical
image communications field.  Other products which are available do trade
off bits of digital data and/or are variations of amateur slow scan
television techniques.

     There are manufacturers (such as Raytel) who are staunch supporters of
data compression of television images. Most still frame TV systems using
data compression use the discrete cosine transform, which I have found
to completely destroy the image where the contrastural information is
most subtle and highest.  Nevertheless, there are some applications where
data compression is acceptable.

     Another temporal compression scheme which uses "telephone lines" (in
a sense) involves the 56/64kbps DDS now emerging in the United States. You
can also get these kinds of circuits (I think) in SCPC form from some satellite
time brokers and owners. Widcom Corporation is the largest proponent of 
live, temporo-spatial compressed NTSC television. Their equipment has the
smarts to not refresh stationary parts of the image, and thus is "semi-live"
in that only the parts of the image needed to convey understanding are
refreshed when they move or change hue/saturation/luminance beyond certain
limits. In still television these equipments are quite good, and even for
certain teleconferencing applications, the image quality is acceptable.
You wouldn't want to telecast "Live from Studio 8-H" or the "Hallmark Hall
of Fame" with it, though.  This system actually works, and has applications
for people who can stomach the tradeoffs involved.

     Hopefully, the AI people and neuropsychiatric types will discover something
about human vision which can be implemented in hardware. Until the biology
and mathematics about human vision is fully understood, you'll be stuck with
raster scan television that can't be transmitted live through a 3000 Hz
C message circuit.

York David Anthony
Director of Engineering
DataSpan, Inc