[comp.misc] Information Density of Sight and Sound

RMC100@psuvm.psu.edu (Randy Carraghan) (02/27/90)

I just finished downloading some GIF files and noticed that the simplest
images, the cartoons, required about 10K for a single frame.  Assuming
we play the cartoon at 20 frames per second, approximately 330 megabytes
of disk space would be required to store a 30 minute cartoon.  This would
require, of course, that the computer be able to read the images off the
drive and display the images quickly enough to simulate an animated cartoon,
which is far from being possible on my IBM-PC AT.  The full-screen, color
images used on the order of 200K bytes of disk space for a single frame,
and these images used only 256 colors total.  Even using CDs, you'd only be
able to store a minute or so of animation.  Now if you were to implement some
sort of compression routine, you'd be able to increase the number of frames
you could store, but the playback speed would become slower than it already
is.  How is it that two hour movies are stored on a single video disk?
(Granted, video disks are larger than standard CDs, but the ration wouldn't
account for the 100-fold storage difference).  What kind of hardware
(computer and disk) capabilities would be needed to process video (or just
sound) at real time speed?  Another way of asking this is, how many bytes
per second would need to be processed (read in, transformed, displayed) for
various media types (simple cartoon, b/w, color, AM/FM/CD quality sound only).

Thanks for any info!

Randy Carraghan (rmc100@psuvm.psu.edu)

fwb@demon.siemens.com (Frederic W. Brehm) (02/27/90)

Randy Carraghan (RMC100@psuvm.psu.edu) asks:
>... Even using CDs, you'd only be
>able to store a minute or so of animation.  Now if you were to implement some
>sort of compression routine, you'd be able to increase the number of frames
>you could store, but the playback speed would become slower than it already
>is.  How is it that two hour movies are stored on a single video disk?

Intel's DVI (invented at David Sarnoff Research Center when RCA was still
around) achieves a 160 to 1 compression ratio for full motion video (TV
quality).  This allows a full hour of video to be put on a CD.  Intel has
some plug-in boards for AT computers that allow playback (decompression) in
real-time.  Compression could not be done in real-time, the last I heard.
The latest Electronics magazine has articles on "Multi-media" systems,
including DVI, CD-I, and Apple's systems.

The video disk is the name given by RCA to their disk.  It could store one
hour of uncompressed video.  The laser disk (Phillip's video disk) stores
uncompressed video, too.  It has several different modes that trade time
for random access or other capabilities, but it does not use image
compression.

Both of these video disks store images in an analog form, not digital.
Their bandwidth is sufficient for storing analog TV images (because the
disks were designed that way!).  Errors in the analog world are called
"noise".  People are used to seeing some noise on their TV screen, so the
disks were not designed with analog error detection or correction (besides,
how would you do that?).  The analog disks were designed to have
"acceptable" (as defined by someone) noise levels.  Their digital
counterparts have error detection and correction to provide a nearly
error-free (noisless) digital signal.

An analog signal converted to digital and then modulated for transmission
over an analog channel requires more bandwidth than the original analog
signal.  Add error detection and correction to the digital signal and you
need even greater channel bandwidth.  This is why your calculations would
lead you to believe that the laser disk can't work at full speed!

Fred
--
Frederic W. Brehm	Siemens Corporate Research	Princeton, NJ
fwb@demon.siemens.com	-or-	princeton!siemens!demon!fwb

ejp@bohra.cpg.oz (Esmond Pitt) (02/28/90)

In article <24837@siemens.siemens.com> fwb@demon.UUCP (Frederic W. Brehm) writes:
> Errors in the analog world are called "noise".

No. Errors in the analogue world are called 'distortion' if they are a
function of the signal, and 'noise' if they are independent of the
signal.


-- 
Esmond Pitt, Computer Power Group
ejp@bohra.cpg.oz