Posted by Jan Panteltje on 10/11/06 11:42

On a sunny day (Wed, 11 Oct 2006 11:47:55 +0100) it happened Roderick Stewart
<escapetime@removethisbit.beeb.net> wrote in
<4aipi25oivv0ilp6fm2pohtsuad8drhv69@4ax.com>:

>On Wed, 11 Oct 2006 03:48:36 +0300, "Jukka Aho" <jukka.aho@iki.fi>
>wrote:
>
>>> "Compatible" color is called that because it is built on the base of
>>> the original black&white video image. If you know that
>>> "white" is made of 11% blue and 30% red, you can calculate
>>> the green as the remainder. Note that this is the proportion
>>> for NTSC, dunno that PAL/SECAM are exactly the same?
>>
>>PAL uses YUV, NTSC uses YIQ, and SECAM uses YDbDr.
>
>These are effectively different notation for the same things, i.e.
>luminance and two colour-difference signals based on R-Y and B-Y.
>
>The one minor qualification is that broadcast NTSC is supposed to be
>encoded with a higher chrominance bandwidth along one encoding axis
>which doesn't exactly align with either of the two colour-difference
>axes, but I understand this is often ignored by TV receiver designers
>and nobody notices. You can demodulate an NTSC signal in R-Y and B-Y
>phases (instead of I and Q phases) and matrix the resultant signals
>with Y to produce R G and B in exactly the same way as for PAL The R-Y
>and B-Y signals from a SECAM decoder will be exactly the same too.
>Signals are often transcoded between all three standards without any
>change to the R-Y and B-Y derived signals other than to adjust their
>amplitudes to conform to the appropriate standards.
>
>Rod.

There are a few basic differences between UV and IQ, here it is described
rather well:
http://en.wikipedia.org/wiki/YIQ
In NTSC the vectors are rotated 33 degrees.
As for representation of where all the colors are relative to R-Y and B-Y see
the 'vectorscope':
ftp://panteltje.com/pub/vectorscope.gif.

The important thing with PAL is however that every line 'V' flips 180 degrees.
The reason for this is that then, when you add the color information in 2 subsequent
lines, any phase error will cancel (at the expense of a slight saturation error,
and some loss of vertical color resolution).
To be able to add the color data in the current line to the previous one,
the previous one is stored in a 64 uS glass delay line.

Here is some info:
http://www.ee.washington.edu/conselec/CE/kuhn/ntsc/95x4.htm
The line by line flipping of V needs to be synced, this sync signal is half the H frequency
and derived from the now 90 degrees swinging burst.
This is a bit easier to see then in the above link, if you flip the V modulator 180 degrees,
automatically the burst will flip (as it is a sum signal in U and V) 90 degrees with it.

Cheap quality TV sets (Sony did this once to avoid paying for the PAL TV patent) use no delay
line, and leave the integration of the color error to the eye.
Sony used a NTSC demodulator, simply rotated the axis, only PAL TV with hue control ;-)
PAL has normally no hue control.

Anyways, soon all will be digital, and these techniques that were quite inventive forgotten.

[Back to original message]