Re: [AV Media Matters] Hi-Def Film Transfers

[dave.bancroft@philips.com]

(MET)
          (envelope-from dave.bancroft@philips.com)
Received: from smtprelay-eur1.philips.com(130.139.36.3) by
gw-nl4.philips.com via mwrap (4.0a)
	id xma004338; Sun, 29 Oct 00 11:36:46 +0100
Received: from notessmtp-nl1.philips.com (notessmtp-nl1.philips.com
[130.139.36.10])
	by smtprelay-nl1.philips.com (8.9.3/8.8.5-1.2.2m-19990317) with ESMTP id
LAA03607
	for <AV-Media-Matters@topica.com>; Sun, 29 Oct 2000 11:36:45 +0100 (MET)
Received: from EMLMS01.DIAMOND.PHILIPS.COM (emlms01sv1.diamond.philips.com
[130.143.165.213])
	by notessmtp-nl1.philips.com (8.9.3/8.8.5-1.2.2m-19990317) with ESMTP id
LAA28912
	for <AV-Media-Matters@topica.com>; Sun, 29 Oct 2000 11:36:45 +0100 (MET)
Received: by EMLMS01.DIAMOND.PHILIPS.COM (Soft-Switch LMS 4.0) with snapi
          via EMEA1 id 0056900013301336; Sun, 29 Oct 2000 11:45:14 +0100
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1; name="MEMO 10/29/00 11:36:36"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Dear Todd,

I would approach this from the point of view of efficient use of
transmission bandwidth: presumably you will be doing HDTV transfers
because an HDTV transmission opportunity is occurring. This will
afford the chance to show more information in the content than would
a standard-definition transmission, provided there actually is
something more in the content to show.

That proviso is crucial, because otherwise the extra bandwidth would
be wasted.

The answer hinges on the following:

1) How good was the original film before it was transferred to the
existing 1" or SP? - I mean in terms of resolution and graininess.

2) How much of the wanted information did the transfer take away,
i.e. reduction in resolution through analog recording bandwidth
(poorer in SP than in 1")  ?

3) How much unwanted "junk" did the transfer add to the content,
e.g. noise, and moire (1") or noise on edges (Beta SP) ?

The answers to these three questions will tell you what you are
starting with before you attempt to upconvert to HDTV. An
upconversion from a good digital standard-def master could be easier
than from analog, but if there wasn't that much information in the
original film (e.g. standard 16mm), it might not have made that much
difference. But what you will be up against with analog tape
playback is the fundamental issue that "sharpening up" edges in the
wanted image will also enhance the visibility of all the "junk" -
the noise and moire (FM beat frequencies). To some extent that can
be mitigated in the noise area with careful use of a modern noise
reducer, but it gets more difficult with moire; anti-moire filters
can soften things up, which takes you back where you started from.

And to deal with the wanted information itself, of sharpening up
edges in the original to make it look like "HD:" There is a
difference between 16mm to 35mm blowups and standard def to high def
blowups. In the first case, there is no opportunity for "sharpening"
(in fact there is softening caused by the optical MTF of the
printer, which is greater in an enlargement printer than in a
contact printer) and the only significant "artifact" is the
magnification of the 16mm grain. Also the MTF curve in the film,
both in the 16mm original and in the 35 mm print, is a gently
rolling off curve.

Contrast this with video: all recorded and transmitted video signals
will have been subjected to a specific form of filtering necessary
to prevent aliasing. In digital it's to prevent aliasing in relation
to a sampling clock; in analog it's to prevent aliasing in relation
to some form of modulation (e.g. preventing or minimising moire in
an FM recording system). This makes video MTF curves have a flattish
response followed by a precipitous plunge, compared to film. This is
in order to get as much "sharpness" as possible, short of aliasing,
in a highly-restricted bandwidth (even in HDTV, the principle is the
same).

Such MTFs behave quite differently from film-type MTFs in blow-ups,
with the risk of nasty-looking black edges around objects, even
though there might not be any real "resolution" (fine detail)
present. You should look to see how much aperture correction
(sharpening) was already applied in the standard-def transfer. If
there was already a lot of it, it will not be a very co-operative
candidate for upconversion. Unfortunately, it's precisely SDTV
transfers that are likely to have lots of aperture correction,
because of the natural desire to try to create sharpness where it's
lacking (because of SDTV bandwidth) - video engineers like sharp
pictures!

Finally, you have to think about where the finished HD upconversion
is going.  The only high-definition distribution outlet that does
not involve compression is recording back onto film again, which I
suspect is not your objective. All electronic HD distribution
involves compression. In the case of DTV, it means getting about a
Gigabit per second of information down to about 19 Megabits per
second. This is aggressive compression! Compressors like clean
sources - they don't like noise and moire (especially noise) because
it looks like unpredictable information that gobbles up the
available bandwidth leaving less room for the real information,
which then gets compressed even more aggressively (read: MPEG
blockiness). So what compressors do with such difficult material is
to prefilter it!  They will use intelligent noise reduction to the
extent possible (or sometimes beyond the extent possible!) and when
that isn't enough, they will roll off the frequency response,
sometimes to a lower value than that associated with standard
definition!. This is simply to give the compressor fewer bits to
compress!

Bottom line:

a) Think hard about whether you really have something to fill the HD
bandwidth with - in the original material and in any upconversion.
If not - be green and save bandwidth!

b) In my book, I'd rather see soft but clean images, than "enhanced"
cartoon-like images crawling with blackened edges and jaggies and
erupting with MPEG blockies - especially on a big display - but do a
test and see what you think (use a big, really-high-resolution
display).

c) The only advantage HDTV might have in the case of
non-high-resolution source material is that delivery to the home
consumer occurs without the classic NTSC analog transmission
artifacts (mostly noise and multipath).

Bottom-bottom line:

You don't still have the original 16m and 35mm films do you? A fresh
HDTV transfer from them would be another matter altogether.

I didn't answer your last question about the choice of future tape
format.  I'd like to invite Jim Wheeler to answer that one, because
he has devoted a lot of thought and research to it.

Regards,

Dave Bancroft
Business & Technology Development Manager, Film Imaging
Philips Broadcast

twieneke@mpimedia.com on 28/10/2000 21:20:16
Please respond to AV-Media-Matters@topica.com@SMTP
To:	AV-Media-Matters@topica.com@SMTP
cc:
Subject:	[AV Media Matters] Hi-Def Film Transfers
Classification:=09

     As a media archivist with a commercial stock footage archive,
we are constantly faced with Darwinian adaption to the ever-changing
demands of technology and the producers who prompulgate them. At
present, we are trying to deal with HDTV, and are subsequently
caught in the quagmire of confusion that seems to come with it.

     Much of our 16mm and 35mm film tranfers have been mastered to
analogue formats ranging from 1" to Beta SP, so it seems that they
are essentially useless in terms of conversion to hi-def. Is this a
correct assumption? I understand that a hi-def conversion is similar
to a 16mm to 35mm blowup, only with less resolution, so what might
happen if a 1" or Beta SP tranfer of a 16mm film was to go through a
conversion? Are there any technological solutions to this apparent
degradation of image? And as for our future film transfers, is
Digital Beta the most appropriate format to master onto in the event
that it might be converted to hi-def in the future (telecine
transfer to hi-def are too cost-prohibitive for us at the moment)?

     Thank you so much for any and all help in this matter!

todd wieneke
MPI Media Group