[rec.video] MacII and video editors and compressors, is 48 to 1 possible?

jrg@Apple.COM (John R. Galloway Jr.) (03/26/90)

	Recently I saw on some PBS computer show (The Computer Cronicles I
think) an issue on desktop video.  They had a piece on a very high end
MacII based video editor (about $80K) though no mention of what extra hw
it was using (and I do not remember the manufacturer).  You read in your
tape (looked like std VHS) did all the editing from disk, then made a new
tape.  The claim was that it took 600MB of disk to store 20 mins of video.
At 30 512 by 512 by 24 (full) frames per second thats only 16KB per (full)
frame or a compression ratio of about 48 to 1.  Just for the editing process
you could drop every other line, and then known 20-25 to 1 compression parts
would make it, but to produce the final tape you would want full info, which
also would have to come from disk I would think.  Is 48 to 1 possible?  Is this
likely just false information?  Does anywone know the name of the manufacturer
of this beast?
	On the same show (FYI) another company displayed a MicroChanel board
that did real time full color video (with sound) display in a MSWindows window,
including real time scaling/interpolation to ANY size.  All with no host cpu
interaction, other than window/mouse actions.  Very impressive.  So when is a
MacII version comming?
-- 
internet   jrg@apple.com      John R. Galloway, Jr.
applelink  d3413              CEO..receptionist
human     (408) 259-2490      Galloway Research

These are my views, NOT Apple's, I am a GUEST here, not an employee!!

mxmora@unix.SRI.COM (Matt Mora) (03/27/90)

In article <39830@apple.Apple.COM> jrg@Apple.COM (John R. Galloway Jr.) writes:
>
>	Recently I saw on some PBS computer show (The Computer Cronicles I
>think) an issue on desktop video.  They had a piece on a very high end
>MacII based video editor (about $80K) though no mention of what extra hw
>it was using (and I do not remember the manufacturer).  You read in your
>tape (looked like std VHS) did all the editing from disk, then made a new
>tape.  The claim was that it took 600MB of disk to store 20 mins of video.
>At 30 512 by 512 by 24 (full) frames per second thats only 16KB per (full)
>frame or a compression ratio of about 48 to 1.  Just for the editing process
>you could drop every other line, and then known 20-25 to 1 compression parts
>would make it, but to produce the final tape you would want full info, which
>also would have to come from disk I would think.  Is 48 to 1 possible? 

I seen the same show but only the last part about the IBM microchanel.

A company called C-Cube as a chip that can do up to 200 to 1 compression.
Its based on the JPEG standard. The chip is called CL550. Compression 
time is less than two seconds but mostly will be less than .25 sec.

The current issue of macweek has an article about the chip.
(MacWEEK 27 March 1990) vol 4 no. 12







-- 
___________________________________________________________
Matthew Mora
SRI International                       mxmora@unix.sri.com
___________________________________________________________

hsu@eng.umd.edu (Dave "bd" Hsu) (03/27/90)

In article <39830@apple.Apple.COM> jrg@Apple.COM (John R. Galloway Jr.) writes:
>	Recently I saw on some PBS computer show (The Computer Cronicles I
>think) an issue on desktop video.  They had a piece on a very high end
>MacII based video editor (about $80K) though no mention of what extra hw
>it was using (and I do not remember the manufacturer).  You read in your
>tape (looked like std VHS) did all the editing from disk, then made a new

This sounds a lot like the Avid Technologies edit system.  They use a
modular deck controller system so you have your choice of source and
target decks, although U-Matic seems to be their favorite.

> The claim was that it took 600MB of disk to store 20 mins of video.
>At 30 512 by 512 by 24 (full) frames per second thats only 16KB per (full)
>frame or a compression ratio of about 48 to 1.

Except that this is a _video_ editor, not a film editor.  30 fps, not 24.
You can't reasonably expect 512x512 anyway from NTSC video, and you don't
need it for editing anyway; a field freeze is sufficient.  Let's say that
100x75 is the minimum tolerable size for a display window.  As nice as 24
bits may be, you can get by okay with 8 or so.  Multiply it out, and you
get 44 minutes of cheezy, uncompressed video on a 600Mb drive, eating
225kb/sec.

Now look at your limiting factors in a stock Mac II.  You can realistically
get a little better than one Mb/sec off the disk without clobbering the OS.
A Mac IIx grinds along at a couple MIPS, so you'll have no trouble getting
data to the screen, but there really isn't the power to do much decompression.
With some sensible buffering to cover up disk seeks, you can safely fly about
500kb/sec of raw video data, enough room to run two 145x110 display windows
at a time and two 10kHz sample rate 8 bit audio channels to boot, without
using any compression and without adding any hardware.

Back to your original question, I recall that Sarnoff Labs had announced
an experimental video compression chipset as part of their CD-I stuff in
late '86 which achieved between 50:1 and 200:1 compression ratios.  No
idea how well it worked, nor how good it looked.

-dave

--
Dave Hsu	 Systems Research Center, Building 115    (301) 454 8867
hsu@eng.umd.edu  The Maryversity of Uniland, College Park, MD 20742-3311

"I'm fishing.  No I'm not, I'm newting!"  - A. A. Milne

alex@grian.cps.altadena.ca.us (Alex Pournelle) (03/28/90)

jrg@Apple.COM (John R. Galloway Jr.) writes:

>	On the same show (FYI) another company displayed a MicroChanel board
>that did real time full color video (with sound) display in a MSWindows window,
>including real time scaling/interpolation to ANY size.  All with no host cpu
>interaction, other than window/mouse actions.  Very impressive.  So when is a
>MacII version comming?

VideoLogic makes an ISA and MCA version of the 4000 currently; they are
in Real Soon Now on the NuBus one.

This is quite a card, sports-fans.  2 separate (gen-locked) sources,
plus native computer graphics, takes PAL, NTSC, SECAM (!) or RGB (Brit
firm, so standards means "worldwide") and interpolates on the fly.
$2500 SRP.  Woo.

On the ISA system, by the bye, they snarf the VGA signals from the VGA
feature connector on the top of the board.  I presume that the MCA one
uses the MCA video bus for the same thing.

	Alex

boissier@irisa.fr (franck boissiere) (03/28/90)

From article <1990Mar28.104021.2521@grian.cps.altadena.ca.us>, by alex@grian.cps.altadena.ca.us (Alex Pournelle):
> jrg@Apple.COM (John R. Galloway Jr.) writes:
> 
>>	On the same show (FYI) another company displayed a MicroChanel board
>>that did real time full color video (with sound) display in a MSWindows window,
>>including real time scaling/interpolation to ANY size.  All with no host cpu
>>interaction, other than window/mouse actions.  Very impressive.  So when is a
>>MacII version comming?
> 
> VideoLogic makes an ISA and MCA version of the 4000 currently; they are
> in Real Soon Now on the NuBus one.
> 

Who cares about a card when RGB technology manufactures an external device
(no slot used) which does the same thing and is compatible with about
all 19' screens (Mac, PC, Sun, HP Appolo, Dec, etc ...) ......

Of course you cannot do image processing.
--
Franck BOISSIERE                        boissier@irisa.irisa.fr
Prototyping Lab Manager                 boissier@ccettix.UUCP
C.C.E.T.T.   B.P. 59                    boissier%irisa.irisa.fr@uunet.uu.net
35512 CESSON SEVIGNE CEDEX  FRANCE    

milo@cartan.math.nd.edu (Greg Corson) (03/29/90)

I think if you take a closer look at that video editing system on the
computer cronicals, you'll notice they were running at considerablly less
than full video resolution.  To me, the images looked like they might be
on the order of 128x128 or so.  In full color they look good enough for editing
use and require a fairly small amount of memory.

I'll wager the company just uses these little images, then once the data
is setup they use the Mac to control a standard A-B roll editing system
to make the final cuts.

Greg Corson
19141 Summers Drive
South Bend, IN 46637
(219) 277-5306
{uunet, rutgers}!iuvax!ndmath!milo
milo@ndmath
GEnie:  GCORSON