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

Re: i/ Luma coefficients for DTV



At 4:46 PM -0400 8/1/98, Charles Poynton wrote:
>Someone commented that the issue of luma coefficients for HDTV can be
>addressed by "tagging" the data, and that in fact such tags are implemented
>within the MPEG-2 stream. MPEG-2 does provide tags. Some people call this
>"metadata."
>
>But think of the bigger issue.
>
>Consider the lowly light switch. It is mounted to the wall box using two
>screws. In North America, these screws are at 2.375 inch centers. That's a
>standard. When I buy a light switch, it fits.

This is a complete hollow argument. We are not talking about a system where
every component must behave the same from acquisition to display...we've
been riding that horse for fifty years. Rigid standards for light switches
and light bulbs, both the kind that glow white and the kind that make
pictures are part of the tight coupling required in an analog world...Where
point standards make sense and enable markets, like the light switch
example, the marketplace will support them. Where they are meaningless,
imposed mainly to protect entrenched interests, the marketplace will go
around them.

Over the past week, I and others have made a rather compelling case for the
FACT that we now have the ability to manage colorimetry, and to calibrate
devices that operate in different color spaces. The problem that Charles
cites with 601 and 709 colorimetry is a trivial task for a color management
system to deal with. I'm not saying that it's a good idea that these should
be different, but I certainly am not going to lose sleep over something
that is so easy to deal with. I'm much more concerned about the huge color
gamut that is available on most displays that we simply throw away in the
world of video...now that we have image compression systems that give us
precise control over every parameter of the image, I want to stop the
lunacy and start preserving this valuable color information that we are
simply throwing away.

>But the deeper problem is that there is no functional advantage to the new
>standard. It might be desirable to decrease the screw centers by a factor of
>two, to make light switches more compact. Or to establish a new standard
>with much wider spacing, to make switch mounting more stable. But a small
>change, from 2.375 to 2.5 inches, is just a nuisance. Tagging does not
>adress the larger problem, which is the lack of a functional advantage.

In this example Charles is correct. Why change something that is working
for no realizable advantage...whoops, that sounds like what's happening
with DTV :-)

But whenit comes to delivering better mostill and moving pictures, one can
make a big case for change...fundamental change that will deliver real
benefits. So let'n stop arguing about the piddling cdifferences between 601
and 7009 colorimetry and start talking about a quantum leap in the
colorimetry we can deliver to everyone.

>HDTV standards were initially adopted with non-square pixels. There is an
>important long-term functional advantage in square pixels - namely, ease
>in exchanging image data with other industries. That aspect of HDTV
>standards was changed - HDTV standards now have square pixels. Adopting
>HDTV with altered luma coefficients would carry a cost comparable to having
>square and non-square versions of HDTV studio standards.

I've been fighting for square pixels for a long time, but in rreality we
ccan live with non=square pixels in some areas of the system...as long as
we know what we are doing. By the way, the cost of dealing with square and
non-square pixels is being adsorbed into every DTV receiver and set-top
box, so that we can deal with 18/36 formats. If we extend Charles argument
to it's logical conclusion, we must argue for a DTV system with only one
format, one colorimetery, one luma equation...etc.

>Metadata would not have resolved the 60 Hz vs. 59.94 Hz issue, and would
>not have resolved square vs. nonsquare pixels. Metadata can describe the
>parameters, but it cannot fix them! Metadata will not resolve a change in
>luma coefficients, and the change carries no functional advantage. SMPTE
>and ATSC should adopt a recommended practrice calling for the Rec. 601 luma
>coefficients to be used in all forms of DTV, ATV, and HDTV.

SMPTE should not waste time on this. Instead, they should be developing an
infrastructure where we can support any colorimetry, including full
bandwidth RGB and extended range luma (10-14 bits log), so that we can
interoperate with film and other high resolution imaging systems.

Regards
Craig





---
Thanks to Filmworkers Club Dallas for support in 1998.
No product marketing allowed on the main TIG.  Contact rob at alegria.com
1000 subscribers in 39 countries on Sun Aug  2 11:14:25 PDT 1998 
subscribe/unsubscribe with that Subject: to telecine-request at alegria.com
complete information on the TIG website http://www.alegria.com/tig3/