[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

FW: Spectrum Fees Hard To Grock in a DTV World

The message forwarded below primarily concerns DTV program transmissions
which have been originated electronically (via TV camera) in an
interlaced form.  Many might question the relevance to TIG.    I believe
it is twofold:

1) It reinforces the fact that unless any inherent 3:2 sequence is
precisely managed and eventually inverse-Telecined out of the program
immediately prior to compression and transmission, less efficient and
lower quality DTV transmission will result.   The downstream effects of
these problem areas in next year's DTV compression/transmission world
have not been widely communicated (especially in other forums).   Is
there really anyone working on a 24 frame per second progressive VTR/VCR
for the professional market?  Will this have SMPTE292M connectors on the
back?  A new variation of 294?  Inquiring minds need to know.

2) It helps explain why the 24 frame/sec progressive output of the TIG
talent pool promises to be the highest quality source of DTV program
content delivered to the American consumer starting next year and for
many many years to come.   Higher definition television cameras will
certainly change the way some business is done.  But I believe the
overall demand for film-originated content will escalate, especially due
to the superior nature of the delivered DTV experience enabled by the
highly flexible ATSC transmission standard when broadcast from a 24P
production medium.  (Whether that be 480/24P, 720/24P or 1080/24P.)

Mark Schubin had some very good comments regarding my original message
below.  I agree with most of them and would encourage him to jump in
here too so this list can benefit.

> -----Original Message-----
> From:	Tom McMahon 
> Sent:	Saturday, October 11, 1997 2:35 PM
> To:	'OpenDTV at pcube.com'
> Cc:	Tom McMahon
> Subject:	Spectrum Fees Hard To Grock in a DTV World
> When NTSC was the only game in town the notion of a TV program's
> analog channel spectrum requirements were fairly well understood.
> However, with the proliferation of potential DTV broadcast formats as
> well as the flexibility of MPEG-2 compression technology, the notion
> of spectrum utilization for any random TV "program" or "channel", or
> multiplex thereof, becomes almost meaningless.   This is especially
> true unless one defines some standardized metric of delivered program
> quality at a given bitrate and some expected nominal screen size upon
> which you intend that quality to be delivered to an Average American
> Viewer-Unit.
> The ATSC's Table 3 contains a flattened N-dimensional production
> format space with equivalent pixel rates (information rates) spanning
> almost a factor of ten to one.  The lower resolution production
> formats have less information per second than the higher information
> (definition) formats.   1920 by 1080 at 30 interlaced frames per
> second has about the same amount of information per second as 1280 by
> 720/60 progressive, more or less.  704 by 480 acquired at 60 frames
> per second is about a third that.  The popular equivalent of our 601
> format is somewhere between a 5th and a 6th of the higher-end formats.
> The bottom end format (which may never actually see the light of day,
> 640 by 480/24 frame/sec) is approximately one tenth.  But those
> comparisons only hold true in the production domain.  Once these Table
> 3 formats become MPEG-2 compressed, as they must for ATSC
> transmission, comparisons of net delivered information rates become
> very difficult to compare, and even harder to tax meaningfully.
> Invoking the "H" word into the discussion of DTV becomes similarly
> meaningless unless one specifies a standardized metric of delivered
> program quality at a given bitrate and some expected nominal screen
> size.  Simply because a program was produced in a studio using one set
> of numbers doesn't mean that same information rate will be delivered
> to an average viewer at a better perceived quality compared with any
> other set of numerology.  Making the receiver's screen bigger doesn't
> make it measurably better, although technologies such as line doublers
> can improved perceived results by eliminating line structure across
> some range of screen sizes.  (But line doublers are with us to stay
> whether we broadcast DTV or not!)
> The fact that all of these production formats end up getting funneled
> through a pipe that's as much as 50 times smaller than the information
> rate in the production domain means something has to give.   What gets
> thrown away is sometimes quite visible to the average viewer.  There
> is no DTV Santa Claus.  What ends up going through the "spectrum"
> bears little relationship to picture sampling numerology anyway as it
> is statistically filtered spatial frequency information, not "pixels".
> (There are NO "scan lines" in the broadcast channel!)   As with any
> large and complex system, engineering tradeoffs sometimes create
> better results with one approach to the system design than with
> another.  Our understanding of those tradeoffs will only get better
> with time.   Now is not the time to cast them in stone; to do so would
> be saying that compression technology is standing still.
> Recent compression tests of one of the popular ATSC Table 3 formats
> (480/60P) produced some interesting results.  These tests were
> performed using 32" direct-view CRT screens, which certainly lie
> somewhere in the sweet spot for the sales volume of American TV sets
> for decades to come.  Compressed at a nominal 20 Mbits/sec, the
> program was indistinguishable from original footage.  At the 10 Mbit
> level artifacts became discernable to some expert viewers but the
> program generally remained stunning.  At 6 Mbits artifacts were
> apparent to some viewers during high stress footage, even at several
> pictures heights of viewing distance.    Certainly one basis condition
> for any definition of High Definition should be the degree to which a
> particular set of system engineering tradeoffs is free from visible
> compression artifacts, no?  In comparison, footage created at the
> popular ATSC Table 3 format of 1080/30 interlaced has noticeable
> artifacts even at 20 Mbits on the same screen size.  At 10 Mbits I
> myself find this 1080/30I content is unwatchable.  (Actually this
> footage is 1035/30I for obvious reasons, but that's yet another
> story...we all use the term 1080/30I somewhat loosely.)
> Broadcasting a signal that is not directly compressed from 1080/30I
> does not mean the delivered signal to the American public is not HDTV.
> And, it certainly does NOT necessarily mean a channel multiplex.  The
> most cost effective utilization of the allocated DTV spectrum is to
> use it as leverage to help the US broadcasters Get Digital Now.  We
> must help stimulate a market for both digital content and a broad DTV
> receiver base.  Moore's law and the open market will sort everything
> else out a relatively short time thereafter, certainly in less time
> and with far more room for innovation than our analog broadcast
> television heritage left open to us.
> For an equivalent delivered viewing experience on a 32 inch screen,
> 1080i is less efficient and requires a higher bitrate through the
> channel than may of the other formats.  This is precisely analogous to
> the situation we were in several decades ago with the state of
> automobile engineering in this country.  In terms of spectrum
> efficiency and delivered viewing experience to the average American
> consumer, 1080I is a gas guzzler and should be taxed accordingly.
> Incenting the deployment of this particular transmission format as a
> mechanism to fully (artificially) utilize American spectrum during the
> infancy startup phase of DTV broadcasting is a terrible waste of this
> country's natural resources and technological prowess.  As we move
> from an industrialized era to an information era, we'll look back on
> this particular "air pollution" situation years from now as we saw
> smog and Dioxin only a few short decades ago.
> ---
These opinions are solely my own.

Thanks to Dave Best & Post Perfect for support of the TIG in 1997
HDTV discussion thread is at http://www.alegria.com/telecine/hdtv.txt
TIG subscriber count is 867 on Sun Oct 12 05:21:02 PDT 1997
mailinglist digest available.... unsubscribe via a message to
'telecine-request at alegria.com' with Subject: unsubscribe