A guide to the electronic generation of colors
for composite NTSC
© June 27, 1991 by David Broberg
With the latest advancements in electronic memory and computer technology,
sophisticated video graphic systems and character generators are becoming
more powerful. Recently, many video graphic systems and character generators
have been touting the availability of more than 16.7 million colors. Mathematically
this is correct for systems that can generate colors from 256 levels, each
of Red, Green and Blue. (256 X 256 X 256 = 16,777,216). But the limits
of the availability of these colors is sometimes overlooked.
Many products such as television character generators and paint systems
have numerous limits that restrict the use of available colors on a given
object, scan line or the screen. Many times these products use memory saving
techniques such as palettes to allow more colors from less memory. These
are generally hardware limitations that vary from one piece of equipment
to the next and should be addressed by the manufacturers specification
sheets. The intention of this article is to address the true availability
of colors that fit into the NTSC composite specifications. Just how many
of these 16.7 Million colors are valid or legal colors?
First the limits must be established for (composite) NTSC. For the purposes
of this article, I will call a "VALID" color any color that when
properly encoded to NTSC will create a signal which falls within the limits
of 100 IRE on the positive excursion and -20 IRE on the negative excursion,
while not exceeding the 100% saturation limit (S.M.P.T.E. color bars meet
this criteria). By this definition, 100% saturated color bars would be
invalid, as would a "Modulated Ramp" test signal because both
contain excursions that go above 100 IRE on a waveform monitor. The excursions
are measured as the instantaneous sum of the luminance information and
the chrominance information.
While it is true that signals with chroma peaks just over 100 IRE can be
legally transmitted, the reason this level was chosen as the limit of "VALID"
colors is many other pieces of equipment in the video path may not be able
to handle excursions beyond this. VCR's and Time Base Correctors for example,
typically have severe distortion (D.P., D.G., clipping, compression, fold-over,
blooming, etc.) problems with this type of chroma excursion. Chroma excursions
that go below -20 IRE can also cause a disruption in the sync or burst
separation circuits.
For the definition of a "Legal" color, the composite peak excursions
can not go beyond 120 IRE in the positive direction. The modulation level
of a NTSC transmitter is established so that 100 IRE corresponds to 12.5%
modulation (sync tip is at 100%). This means a signal with 120 IRE peaks
develops 0% modulation (effectively turning off the carrier). Needless
to say, signals that go beyond this level cannot be transmitted and are
illegal.
One common method that has been used to determine which colors, from an
electronic video generation device, are valid is to use a waveform monitor
to observe violations. This method would only work if it were practical
to have an engineer looking over the shoulder of the artist, as the artist
created new colors on his palette. If you allow the artist to create freely,
then try to review his work after completion, he will be quite upset that
he must rework his art. It becomes quite frustrating to say some colors
can't be used, while no one can explain which colors may be used. There
is no simple rule that can be used to describe valid colors.
A better way to prevent illegal and invalid colors is to simulate the effects
of an ideal encoder mathematically and test for violations. This could
have the effect of having a simulated engineer looking over the artist's
shoulder, warning if a chosen color might be a problem. Some character
generators and video graphics products have taken the first step in this
process by using a color space more suited to television such as H.S.V.
(Hue Saturation and Value) or H.L.S. (Hue, Lightness and Saturation) instead
of the traditional RGB color space. These color models are often more intuitive
to the artist, but the only trouble with these color models is that they
are not based on the levels created by an NTSC encoder. Even after correction
factors have been taken into account, one significant piece of information
about the signal is still missing, the instantaneous peak levels represented
by the sum of the luminance and the chrominance signals.
A computer model may be constructed to simulate the ideal encoder, once
we further understand the mathematical relationships between RGB and Encoded
NTSC. We begin by deriving the relative luminance portion of the NTSC from
the RGB components:
EY = (0.299*ER + 0.587*EG + 0.114*EB)
Whereas:
EY = The relative voltage potential for the luminance component
after matrixing of the RGB signals, expressed as a percentage of 1, where
1 is the level produced when each of the RGB components are at maximum,
and 0 is set-up level.
ER,G,B = The relative voltage potential for each of the individual
color channels expressed as a percentage of 1, where 1 is the maximum level
that may be generated, and 0 is the minimum level, or set-up. It is assumed
that a gamma correction of 2.2 has been added to each of the RGB signals.
It is important to note, that these levels are a relative percentage of
1. and are proportional to, but do not represent the actual peak-to-peak
voltage levels. This formula also assumes that the gamma correction and
the set-up (pedestal) have already been added to the R,G & B components.
This relative level is handy because any resolution of Digital to Analog
converter (D/A) may be incorporated into the formula by using the chosen
level divided by the number of levels available.
Next the absolute luminance level in IRE units (140 IRE = 1.00 V.) will
be determined, based on the absolute levels of the RGB components.
NTSC ENCODER FORMULAS:
660*EY
+ 53.55
YIRE = -----------------
7.14
EI = 0.596*ER - 0.274*EG
- 0.322*EB
EQ = 0.211*ER - 0.522*EG + 0.311*EB
ESAT = Ö(EI2
+ EQ2)
SATIRE = ESAT * 185.232
PHASE0 = ARCTAN (EI/EQ)+33 (IF EQ
< 0, THEN ADD 180)
IF EI = 0, AND EQ = 0, THEN PHASE = N/A.
IF SATIRE > 100,
OR SATIRE / 2 + YIRE > 100,
OR IF YIRE - SATIRE / 2 < -20
THEN COLOR IS INVALID
TRANSCODER FORMULAS:
ER-Y = (0.701*ER - 0.587*EG - 0.114*EB)
= ESAT * COS(PHASE0)
EB-Y = (0.886*EB - 0.299*ER - 0.587*EG)
= ESAT * SIN(PHASE0)
YBETA = YIRE * 0.00714
R-YBETA = 0.6662 * (ER-Y)
B-YBETA = 0.5271 * (EB-Y)
YMII = EY * 0.7
R-YMII = 0.4618 * (ER-Y)
B-YMII = 0.3654 * (EB-Y)
YMII = The Y signal
used by MII which has no set-up, and is a 7:3 video to sync ratio.
R-YMII = The R-Y color component used by the
MII system. When 100% color bars are used this component will have a 0.6475
Volt P-P level. No set-up is used on the RGB components.
B-YMII = The B-Y color component used by the MII system.
When 100% color bars are used this component will have a 0.6475 Volt P-P
level. No set-up is used on the RGB components.
YBETA = The Y signal used by the BETACAM system. This
Y signal is essentially the same as YIRE, but is expressed in
volts and has sync added at 0.286 Vp-p. Set-up is used and the video to
sync ratio is 10:4.
R-YBETA = The R-Y color component used by the BETACAM
system. When 75% color bars are used this component will have a 0.700 Volt
P-P level. Set-up is used on the RGB components. BETACAM-SP will allow
for a P-P level of 0.934 Volts, which is what is created by 100% saturated
bars with this gain factor.
B-YBETA = The B-Y color component used by the BETACAM
system. When 75% color bars are used this component will have a 0.700 Volt
P-P level. Set-up is used on the RGB components. BETACAM-SP will allow
for a P-P level of 0.934 Volts, which is what is created by 100% saturated
bars with this gain factor.
YSMPTE = YMII
Pb = The SMPTE B-Y color component. The level is
0.700 V p-p, (no set-up) when 100% color bars are applied.
Pr = The SMPTE R-Y color component. The level is 0.700
V p-p, (no set-up) when 100% color bars are applied.