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

Re: 601 versus 709 (formerly YUV/RGB) Color Space Issues



In a message dated 97-12-19 17:09:42 EST, Tim wrote:

<< one is the colorimetry, which some people believe is quite perceptible (and
many don't) >>

This depends to a great deal on the colors themselves.  For what would be
considered "ordinary" TV, such as newscasts and game shows, it is doubtful
that much difference could be seen.  The same is not necessarily so for
animation or scenes with a great deal of color -- particularly in the
highlights and lowlights.

There is an additional factor, which depends on how the YUV is mapped to RGB.
Some computerized devices convert 601 level 16 (which corresponds to 7.5 IRE
in NTSC) to level 0 RGB, and 601 level 235 (100 IRE) to RGB level 255.  This
has the effect of creating a hard black clip at setup and a hard white clip at
100, which has very unfortunate consequences, at least in post production.  

<< typically filtered and then resampled (from 4:4:4 to 4:2:2). One of these
matrices pushes a lot of green information into the "2:2" channels; the low
pass filtering then results in a VERY noticeable loss of resolution when the
imagery is finally reconstructed back up to 4:4:4 RGB space for final display
and consumption. >>

In converting 4:4:4 to 4:2:2, low-pass filtering of the color channels must be
performed because this conversion is effectively subsampling; without the
filtering, higher frequecy components (i.e. detail) would create all sorts of
nasty aliasing problems.  

This is not seen as a defect in television, as the "Y" channel is still full-
bandwidth and in theory carries all the detail information needed to
reconstruct what is generally considered to be an acceptable picture.  It
does, however, cause problems in certain graphics or paintbox applications
where the colors are processed separately and the detail is missed in the
channels that don't have it.  In subjective tests I've performed, I've found
that the difference in images, when viewed 4:2:2 and 4:4:4 on the same monitor
is discernable, but subtle.  

Of course there is no restoration of resolution lost in the first
transformation, but I have not observed any further big losses of resolution
going 4:4:4 to 4:2:2 then back to 4:4:4.  

<< This whole thing is a real mess. >>

I couldn't agree with you more!!!

Best regards,
Christopher Bacon

---
thanks to Andrew Lichstein for support of the TIG in 1998
TIG subscriber count is 910 on Mon Dec 22 07:11:32 PST 1997
complete information on the TIG website http://www.alegria.com/tig3/