[EXPERT TIPS][GUY KUO]
Grayscale vs Color Decoder Problems
"Recently, I've seen several discussions in which novices have attempted to
correct red pushing using red bias and gain controls. I'm posting this in an attempt to
clarify the difference between correcting color decoder problems vs altering the entire
Lets review how color bars are useful in setting and checking color decoding.
Looking at AVIA's Blue Bars, SMPTE Bars, or other 75 color bars one sees a series of
colored bars and a 75% gray bar. The amount of each primary color in each bar is 75% of
maximum. The red bar contains 75% red but none of the other two primaries. Blue contains
75% blue but no green or red. Green contains just 75% green. The cyan, magenta, and yellow
bars each contain 75% of two primaries. Cyan is 75% blue + 75% green. Magenta is red +
blue. Yellow is green + red. The gray bar contains all three primaries each at 75%. The
key message is that each bar is encoded with the same saturation level of each primary
color which is present in that bar.
By examining color bars using only one primary color on screen, one can determine if
color decoding is working properly. Ideally, when viewed in blue only all the blue
containing bars are of identical intensity to each other. In red only, the red containing
bars are also of identical intensity. In green only, the green containing bars are
identical in intensity. Of special importance is the gray bar. It contains no color
difference information for the color decoder to process since it is by definition
colorless. However, its gray is composed of 75% of red, green, and blue. It intensity in
each primary is constant as one adjusts the color decoder's saturation and hue controls.
Hence, it serves as a reference against which the other bars may be checked. Viewing in
single primary is best done by cutting off or covering the other guns. Another way of
doing this is to use red, green, and blue color filters. However a small amount of the
other colors leak through a color filter so it is slightly less ideal. The most accurate
means of checking the bars is to directly sample the waveform of the final video output
stages and see the amplitude of the bars on an oscilloscope. For most people, color
filters are the simplest approach.
Now you may begin to understand why adjusting in blue-only works. You see only the blue
light output of the bars. When you adjust saturation, you alter the amount of color. Hue
alters the distribution of the color. So to set saturation, one pays attention to the pure
blue bar and the 75% gray bar. Since you know that both have the same amount of blue you
adjust saturation to make them both equal in intensity. This works because the saturation
control does not affect the gray bar but does affect the blue bar. Hue alters the
distribution blue relative to the other colors so you look at the bars which contain blue
+ one other primary. That would be the cyan and magenta bars. When hue is correctly set,
those two bars have the same amount of the color blue. The two controls interact so it
takes both hue and saturation adjustments to get all blue, cyan, magenta, and gray bars to
have identical amounts of blue.
SMPTE bars made this easier by putting patches of the comparison color under each bar.
AVIA Blue Bars do that and also add flashing patches of the comparison color to take
advantage of visual sensitivity to flashing and improve viewer accuracy.
That would be the end of the story if color decoders all worked perfectly along NTSC
standards. If a decoder is NTSC standard, setting the hue and saturation perfectly for
blue only would also set it perfectly for making green and red containing bars have
relationships analogous to the blue bars. That is the amount of green in each green
bearing bar would appear equal when viewed in green only. The red bars would also be equal
in red content.
Just like for blue, the pure red bar has the same intensity in red as the gray bar when
saturation is correctly set. The red + one other primary bars (yellow & magenta) would
also match each other in red intensity when hue is correctly set. Similarly, the pure
green bar has the same intensity in green as the gray bar when saturation is correctly
set. The green + one other primary bars (yellow & cyan) would also match each other in
green intensity when hue is correctly set.
You can compare these intensities using standard or SMPTE bars, but the colors to be
compared are not paired with the correct comparison color. That's why AVIA also has
"Red Bars" and "Green Bars." These have their comparison patches
arranged for making comparisons for the red or green bearing colors. The Color Decoder
Check pattern in AVIA is a quick way of checking 75% of all three primaries against 75%
In decoders with "Red Push" setting hue and saturation to make blue bars
appear correctly, sets the decoder to exaggerate red containing colors. You see this in
color bars by looking at the red, blue and gray bars. First look in blue only and verify
that the blue containing bars are all of equal intensity. This double checks that you've
set the decoder for proper decoding of blue. Next, switch to red-only viewing and compare
the red bar vs gray bar intensity. They should be identical to each other. If the red bar
is brighter (in red-only) than the gray bar, then the decoder is exaggerating reds and has
red push. You may wish to use the Red Bars pattern in AVIA to have gray and red physically
together on screen. The Color Decoder Check pattern has patches of varying intensity red
which lets you quantify the error.
Green push is less talked about because it isn't as visually glaring as red push. You
check this with green-only viewing and once more compare the primary color against gray.
The green patch should match the gray bar in green intensity.
Notices that nowhere in this discussion did the grayscale come into play. That's
because it is a separate issue. All three primaries are checked using bars at 75%
intensity. That means all the test points are at the SAME content of each primary. This
eliminates the effect of grayscale imbalance. Imagine that your grayscale was set so badly
that red was 50% higher than it should be. At first glance, you might think this would
throw off your checks for red push. It doesn't because the red bar and the gray bar would
BOTH be altered by the same grayscale imbalance . Both are at 75% red signal intensity and
rise in red output TOGETHER as grayscale red content is boosted. This means you can check
for color decoder accuracy even when gray scale is incorrectly balanced.
Ideally, your display would be adjusted to make gray scale correct AND color decoding
correct. Reducing the amount of red gun output using the grayscale adjustment controls
like bias and gain alters everything which contains any red gun output. That includes the
color of gray. If your decoder has red push and boosts reds by 25% and you reduce red gun
output by 25 percent, you'll get the color red to look right. Unfortunately, that means
everything will be 25% lower in red, including the color of white and grays. It is better
to leave the grayscale correctly at D65 and fix the color decoding axis and gains.