This morning I realized I had made two minor errors in my equations because
I had forgotten to account for black level being 7.5 IRE.*
Y' = ( COLU AND 15 ) * 0.09
hue = 180 degrees - ( COLU / 16 - 1 ) * 24 degrees
saturation = 0 if ( COLU / 16 == 0 ), otherwise saturation = 20 / 92.5
U' = saturation * cos ( hue )
V' = saturation * sin ( hue )
R' = 1.140V' + Y'
G' = Y' - 0.3947U' - 0.5808V'
B' = 2.033U' + Y'
> As with every conversion I've ever seen posted here, the equations yeild
> values below zero and above 1. Do you clip these? Scale them?
Clip the values of R'G'B' to the range 0.0 - 1.0. The reason this occurs is
the TIA generates a simple luma/chroma signal which may fall outside the
normal RGB color cube. So we need to handle it the same way the TV would,
which is to clip the RGB output. This means there is a difference between
the colors (both hue and saturation) at lower luma than at the higher.
To transform non-linear R'G'B' back to linear RGB each component needs to be
put through the following transform (after clipping):
X = X'/4.5 for X' <= 0.081
X = ((X' + 0.099)/1.099)^(1/0.45) for X' > 0.081
(from _Frequently Asked Questions About Gamma_ by Charles Poynton)
* The CX2600 has the BLK pin #6 connected, which (presumably) generates the
required 7.5 IRE step from blank to black. It also uses a 10K resistor for
SYNC and 12K for LM2. The CX2600A doesn't have BLK connected, but both SYNC
and LM2 use 27K resistors. I've reworked the Y calculation based on the
CX2600 values and 92.5 IRE between black and white.
----------------------------------------------------------------------------------------------
Archives (includes files) at http://www.biglist.com/lists/stella/archives/
Unsub & more at http://www.biglist.com/lists/stella/
