RE: [AV Media Matters] compression: film, sp, error concealment/correction, DVTR

[tony gardner <anthony.gardner@cec.eu.int>]

Thanks to Scott Allen for posting his point about film transitions becoming
abrupt on digital betacam but not on betacam SP.
I am at a loss for an explanation on this, especially as digital betacam is
a intraframe compression system. Is it perhaps frame based compression
rather than field based????

That apart, I would not use betacam SP in preference to Digital Betacam for
two reasons:
(1) SP is inherently prone to drop outs which cannot be easily corrected; DB
drop outs are invisible (due to correction). (see below)
(2) SP colour difference signals are compressed more than the TV system
requires - its more a sort of 4:1:1 system than the 4:2:2 uncompressed
video.

Re Jim Lindner's point about error correction and error concealment: of
course he is right about errors and all tape being analogue (in a certain
way of thinking). However I think this is going into details which mask the
broader argument - for this reason I don't think my posting was wrong, only
incomplete!.
But very importantly, Jim's argument should be used to argue for treating
archived video for preservation as files capable of being copies at 100%
accuracy by a IT systems which guarentees it everytime (or tells you it
failed). Nothing less will guarentee the integrity of the archive. I think
this is accepted, but it follows that no existing video format is really a
preservation format just because of this!

Jim specifically states that most of the time digital VT to VT copies are
100% accurate clones: this is correct as he says due to the error correction
implemented in all digital VTRs - it is a fundamental problem but it is
solved 100% accurately so long as the error rate is not too great. This
point must be made strongly (and Jim does near the end).
Error concealment tries to mask errors which are too much for the correction
system, and operates rather like the analogue VT drop out compensators. When
it happens - and it shouldn't with a well maintained machine playing a
properly recorded quality tape - then the digital copy is not a clone, it is
not 100% the original.
Referring back to Scott Allen's decision to use SP, from the point of view
of drop outs: for the same number and type of drop out up to certain density
on an SP and a DB tape: the drop outs will be invisible on the DB
(corrected) whereas on the SP you will see them (despite the Drop Out
compensator). If the number of drop outs increases beyond that (and you are
talking porrly maintained machines or poor quality tape now), then the drop
outs will be getting very noticeable on SP, but will be concealed on the DB,
meaning the picture is not as the original. The effect is probably less
objectionable and nearly invisible except in A/B comparison. Go for even
more drop out and you will see a pitcure from the SP, but the DB will give
up. It will have gone over the well known digital cliff. Thus DB gives good
better pictures for longer, but gives up sooner. (Its the same for other
types of digital machine aswell).
Thats a serious operational problem for anyone using digital machines. The
front panel error indications on a DB are pretty crude - but better than
most other DVTRs.

***But a system for logging error rates on a frame by frame basis is
essential for proper control of the archive: do I have to construct this
myself via the 232 port, or does someone sell the software to do it?***

Tony Gardner
All views and spelling errors are my own.
EC AV Service, Brussels.

Moderators Comment:
Some excellent points. Yes I would argue for a preservation standard that
had 100% error correction as one of many elements of a specification.
Another would be the PROVEN ability to reconstruct an image when error
correction has failed for one reason or the other. Do we have that in ANY
digital format now? I don't think so. There are many other things that would
be on my list also - these are just a couple of points.

These days it is very unfashionable to actually point out some of the
ADVANTGES of analog formats - but in fact there are many. One of the main
ones is that we are really very good at getting the tapes to play back -
more or less - no matter what happens to them. With all the stones thrown at
2" Quad - the reality is that they have lasted almost 50 years, and we can
pretty much retrieve all of them - sure there are some problems but at least
we can play them back. Can we say this for SURE about digital formats. We
know what happens to tape now with the reduction in RF over time - and that
would inherently increase the noise and make it harder and harder to really
know and correct those errors and once you go over the cliff (and there most
definately IS a cliff) can you get back? Is that risk acceptable? These are
all questons that need to be asked. I say - you don't have to choose one.
Get one of each. The additonal cost of a reel of tape stock is not that
expensive in the scope of things, and it is an EXCELLENT insurance polity on
many levels.

Back to lurking.
jim