Re Frame rate and resolution

Ross Binnie (dianaros@vianet.on.ca)
Tue, 21 Oct 1997 08:34:18 -0400


This is a multi-part message in MIME format.

------=_NextPart_000_01BCDDFC.1F138000
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

Attached is a message from another forum that defines the problems of
sending and receiving images at
high frame rates over the internet or over regular phone lines.

Ross
<dianaros@vianet.on.ca>
------=_NextPart_000_01BCDDFC.1F138000
Content-Type: application/octet-stream; name="Re BW vs. fps.nws"
Content-Transfer-Encoding: quoted-printable
Content-Description: Re BW vs. fps (Internet News Message)
Content-Disposition: attachment; filename="Re BW vs. fps.nws"

Path: =
myth.vianet.on.ca!cdc2.cdc.net!news.maxwell.syr.edu!newsfeed.internetmci.=
com!207.69.200.61!mindspring!news.mindspring.com!usenet
From: sggoodri@mindspring.com (Steven Goodridge)
Newsgroups: comp.dcom.videoconf
Subject: Re: BW vs. fps
Date: Sun, 19 Oct 1997 20:10:31 GMT
Organization: MindSpring Enterprises
Lines: 57
Message-ID: <344a6195.1511709@news.mindspring.com>
References: <34476DC7.64CD@mitdev.com>
Reply-To: sggoodri@mindspring.com
NNTP-Posting-Host: user-37kbn23.dialup.mindspring.com
X-Server-Date: 19 Oct 1997 20:13:14 GMT
X-Newsreader: Forte Free Agent 1.1/32.230

Brent Modzelewski <brent@mitdev.com> wrote:

>I am new to the DVC world. Is there a chart somewhere that correlates=20
>bandwidth, frames/second, processor speed, picture size, color depth.
>
>I have no idea on the quality of service I can expect from video
>conferencing. I see companies claiming frame rates from 4fps to 30fps.

There are two reasons that it's hard to quantify the relationships
between these parameters...which is unfortunate, because it's
important for those of us in the business to clearly explain the
tradeoffs. One problem is that picture quality is often very
subjective, and highly dependent on the type of scene being viewed.
The other is that different transmission and compression technologies
have different properties and are often designed for different
purposes. To really understand the tradeoffs, it's hard to avoid
getting into the nuts and bolts of the technology, _and_ seeing the
actual results on screen to get a subjective perspective.

State-of-the-art broadcast quality digital video, compressed with
MPEG-2, can be had at around 6 Megabits per second. That's 30 frames
per second, full CCIR-601 resolution with minimal artifacts. Granted,
a video conference will compress with less degradation that an action
movie, since motion prediction will be better, but this is a ballpark
figure. (Shameless plug: See http://www.networking.ibm.com/video for
MPEG-2 "education-grade" video conferencing solutions.)

To cut this down to lower bandwidths, we can do one or more of the
following: subsample the picture size to HHR(1/2), CIF(1/4),
QCIF(1/16), etc. before compression, send a smaller number of
frames/second, and increase the amount of compression artifacts due to
the inability to accurately represent the detail or movement in the
picture with the channel bandwidth given.=20

The lower the detail or movement in the image, the less the
compression artifacts will be. Thus a very good looking
videoconference may be performed over a channel unsuitable for an
action movie. MPEG-1 video at 1.5 Mbps over a T-1 line is often
referred to as "VHS-quality" at 30 fps quarter-resolution, even for
high-detail and high-motion scenes. 384 Kbps over three ISDN lines
provides what is commonly referred to "business quality" video. Here
the lower resolution and motion artifacts will be obviously worse than
normal broadcast video, but it is very useful and popular for H.320
conferencing systems used by businesses.

At speeds below a single 64 Kbps ISDN line, the video starts to be
more "novelty quality" than "business quality." Effective visual
communication becomes somewhat hampered by the technology. A single
face centered in the image may look good, but significant moton causes
lots of artifacts. It all comes down to personal taste. In
surveillance applications, for example, high frame rates may not be
important. The best thing to do is try before you buy.

=20
___________________________________
Steven G. Goodridge, Ph.D.
http://sggoodri.home.mindspring.com

------=_NextPart_000_01BCDDFC.1F138000--