[tig] (TIG) HDnet
Thu Apr 1 08:37:34 BST 2004
Peter Swinson wrote:
> I am assuming we are talking HD capture at 1920 x 1080 10 bit Y,Cr,Cb.
> If so then it is my understanding that the Nyquist filters in HD limit the
real resolution to about half the 1920 sampling figure if aliasing is to be
Through my work of the last four years I've come to realize that a lot of
what we thought were good "standards" are no longer. I am now convinced
that we all live in this muck of undefined and less-than-quantifiable
digital processing. Processing that irreparably damages images ... and
everyone seems to happily accept it. Take a look at this, as one small
Mike Orton and I spent some time looking at this problem and taking a few
pictures. The bad news is that, even a six megapixel digital camera --due
to its focal-plane Bayer filter imager-- cannot capture and render images
without distortion. The full impact of the problem is not entirely visible.
Despite that, this is just one sliver of the problem. I've seen inexpensive
devices in the market that, for example, treat PsF material as though it
were interlaced! Impossible you say! Ha! Every single one of you has
precisely this problem in your telecine suites if you are running PsF.
Imagine that, de-interlacing and motion-adaptive algorithms, among other
things, applied to progressive content.
What's happening is that, in the rush to deploy HD to the "masses" designers
count on the fact that HD is inherently hard to evaluate for the
uninitiated. I'll run the risk of insulting a great deal of people on this
list by saying that most of you have NEVER seen real HD. Devoid of that
information, how can we expect anyone at all --much less those without the
technical and empirical background of most on this list- to be able to
evaluate HD vs. anything at all?
Do you KNOW, for example, if that projector --with all of its processing--
is delivering the best and truest HD image? How was each pixel treated from
source to screen? Nobody knows! We happily lash-up a bunch of black boxes
and proclaim that the crud on the screen is HD. Without question. Without
doubt. That's cool for home-theater users, but, that's not what we do,
Yes, there are simple large-area calls that can be made quite easily --like
dynamic range. However, for the most part, no real methodology exists to
separate what constitutes "reference grade" signal processing from "hey,
they'll never see it" processing.
Looking at Bars and Multiburst is just about meaningless any more. I am
starting to believe that we need to establish a standard by which all
digital devices will be measured and evaluated. I am baffled that, for
example, no facility I know of tests signal paths using a zone plate (my
favorite signal for quick system characterization). This is one example of
a signal that would, with one quick look, tell a great deal about the device
it goes through.
There's more, but I'm afraid it'll turn into a bitching session. I'd be
happy to discuss some of this with any of you during NAB.
eCinema Systems, Inc.
martin at ecinemasys.com
ecinema at ieee.org
**** NAB'04 booth SL2756 ****
More information about the Tig