Re: [AV Media Matters] Compression discussion.

[tvmark@gateway.net]

Why is D-1 the format for purists but not D-5?  Both are uncompressed,
and D-5 offers the greater data rate.  Furthermore, its use in HDTV
makes it the more current format.

As for the rest of the discussion, I would agree that compression is not
necessarily evil and has many places.  I often use Digital Betacam and
DVCPRO and have used DCT and Digital-S (to mention some "gentle"
compression video formats) and often use Musicam and MPEG Layer 3 (to
mention some audio compression formats).

I DO, however, have two cautions:

1.  Current presentation technology might not indicate problems.  Sony,
for example, has a demo tape of material shot on HDCAM (4.4:1
compression of reduced-bandwidth high-definition television).  Some of
the material was shot with high-definition lenses; some with standard
lenses.  Shown on Sony's best high-definition direct-view monitor, the
SD-lens material looks close to the HD-lens material.  Seen on the large
projection screen at the Sony Pictures Studios High Definition Center,
however, the SD-lens material sticks out like a sore thumb.
     On the same large screen (fed from an HDIH-3000 projector with
Nikon optics), I could detect differences between the HDD-1000
(uncompressed, open-reel), HDCAM, and the HD D-5 (4:1 compression,
full-bandwidth input to the compressor).

2.  A single round-trip through a compression system does not indicate
problems that may be introduced by concatenated compression.  When we do
a single round-trip through a Layer 3 audio system (128 kbps), for
example, we cannot detect problems with certain source material that we
use.  TWO round trips, however (with the particular compression
parameters in that particular equipment), introduced an audible
"scratchiness" on certain low-level, low-frequency material.  Newer
software has eliminated that particular problem.
     If the sole function of an archive is to provide reproduction on
site, then the concatenated-compression issue need not be considered.
If, however, reproduced material may be subject to additional stages of
transmission or recording compression, then it should be considered.
     Some processes are being developed to help mitigate the
concatenation problem.  Part of the debate over the inclusion of
720-pixel active lines in the U.S. ATSC DTV standard A/53 (instead of
the current 704) relates to just that issue.  Macroblock misalignment
between 704 and 720 prevents simple transcoding between generations.

TTFN,
Mark Schubin
Technological Consultant