Re: [AV Media Matters] Compression

[sallen@gwm.sc.edu]

Well, let's really open this can of worms.

Anthony Gardner and I have exchanged a few nice emails about video
compression and film, analog and digital.  I said, in part:

"After several years of transferring film to Betacam SP, we changed
over to Digi-Beta for several months.  And, I noticed that
transitions which were quite smooth and gradual on the analog Beta,
were abrupt on the digital Beta, to the point of losing visual
information.  I guess, if it's not a 1, it must be a 0.

"I realize that that's an esthetic and not technical point, but as a
colorist, that's part of my job.  The fact that it's a compressed
format (however little), and that I've heard Jim Lindner [Jim,
please correct me if I'm misinterpreting here] even raise a small
question about the validity of the term "digital clone", etc, I just
didn't feel comfortable sticking with Digi-Beta as a mastering
format."

I also used the term "choppy" to describe the way Digi-Beta handles
some film fades.

It would be very helpful to me if anyone with experience on this
list, could comment about the comparative worth of analog Betacam SP
versus Digital Betacam for recording 35mm B&W camera negative.

Neither HDTV or "Hollywood scanning" are within our financial grasp
at this moment...we are an archive, after all 8^).  We're lucky
enough to have an outstanding Phillips FDL60 as a telecine, and a
Betacam SP for video mastering.  We also have a Digital Betacam that
we only use if a client asks for it.

But I'm open to someone persuading me that if we can only afford to
make video masters with an analog or digital Betacam, that digital
is better...

Scott Allen
sallen@gwm.sc.edu
film colorist
Univ. of So. Carolina Newsfilm Library

Moderators Comment:
I am choosing not to comment on several different aspects of the
current thread - I suspect I will at some point - but to Scott's
point. Yes, from a technical point of view the signal that is
actually recorded on the tape is in fact analog even though it is a
"digital tape recording" and this applies to some of Anthony's
comments about a clone being 100% the same. In fact (and I hope Mark
Schubin jumps in here) all magnetic recordings are analog - we live
in an analog world when you get down to it - so digital recordings
are in fact recordings of state changes - you can think of it as a
change of the magnetic field. The assumption is that all of these
changes are correctly interperted by an analog to digital conversion
process -but in fact error is an "expected" part of this process.
Error detection is a standard part of the design of all digital
media recorders for computer data as well as "video data". There are
many "schemes" that are used, and some video tape recorders can
actually tell you the error rate. When the everything is in good
shape the errors normally get corrected - but sometimes this does
not happen and when that occurs the equipment tries to conceal the
error. This is a rather long discussion and there is a great deal
written about this subject in general. UNfortunately Anthony is
incorrect when he mentions that video data is like a computer file.
Wish it were so. Video data when recorded on videotape - lossy or
not, compressed or not, is in fact a sampled stream of information -
not a finite file type. So when copying the informaton or cloning
you are really trying to copy that stream using the same A/D
conversion process from the tape - so error correction and
concealment does in fact become part of the "new" recording whether
it is accurate or not - so it is quite possible for errors to
compound across generations. I might add that this is NOT usually
the case - but it is possible and does happen quite frequently
particularly when the error rate is high. This is one very important
reason to carefully moniter error rates when making digital
preservation masters.

I think that an even more important point, however, relates to
compression and artifacts that occur when different compression
algorithms are used in production. When used in production the video
does need to be uncompressed to perform various activities -
particularly in the area of special effects or when used on a
non-linear editing system that does not use the same compression
type. For instance most Avid systems use MJPEG NOT  MPEG, so the
compression and recompression with different algorithms down the
transmission channel does cause artifacts. On top of this are the
real issues of bitrates and bit depth of any given recording as well
as quantization levels of chrominance. So it really is not as clear
cut as people would have you think.

I am not sure I have provided any "answers" here, but just pointing
out that there are many different layers to this onion!

jim lindner