The Color Bars Signal -- Why and How

Monochrome Video The information to reconstruct a black and white image is relatively easy to understand -- each point in the image is scanned in the familiar line pattern and at each point along the line the camera produces a voltage representing the amount of light at that point. The rapid succession of points scanned -- and the corresponding succession of signal voltages -- creates the video waveform. This waveform is "one dimensional," i.e., only one piece of information is needed for each point in the image. Next: Glossary Table of Contents Color Video -- History The situation gets quite a bit more complex with color video. The problem with color video is it's not one dimensional, but three dimensional -- three pieces of information are needed for each point in the image. There are several ways to convey information about color. Most take advantage of a characteristic of human vision known as the tristimulus theory -- we can satisfactorily reproduce a wide range of color using only three colors of light mixed in certain ways. Color photography and color printing are based on these same ideas -- three filters, three dyes, three inks, etc. (The four-color printing process uses black ink in addition to make dark or gray areas in the image easier to control.) While other combinations are possible, there are some practical reasons for choosing Red, Green and Blue (RGB) as the three "primary" colors for a color video system. In fact, the earliest color television systems used three cameras, scanning in parallel, each with a Red, Green or Blue filter over the lens to produce color video. The display also used multiple images, filters and optics to combine the Red, Green and Blue images into one full-color picture. The Color Bars Signal -- RGB Setting up a system with three cameras, three video transmission paths, and three displays is a chore! Matching the three channels to "track" from black to peak brightness is especially important. Test signals have been developed for this sole purpose. In this RGB system, each video path is normalized so the signals match at black level (zero light) and at peak level (maximum light). A good test signal consists of a square wave with black and peak levels in each of the three channels. If the square waves are timed differently in each channel, all combinations of high and low levels in the channels are exercised (Figure 7-1). Since there are three channels, each with two possible levels, there are two cubed (eight) possibilities. These eight combinations exercise the primary colors individually (Red, Green, and Blue), the secondary colors (combinations of two primaries: Yellow, Cyan, and Magenta), White, and Black. The colors (color bars) are arranged in order of decreasing total brightness: White, Yellow, Cyan, Green, Magenta, Red, Blue, Black. The color bars signal is useful in setting up a RGBcolor video system, and it gives a pleasing image that also conveys some indication of the "health" of the color video system.

Figure 7-1. Video waveforms of the three channels for RGB-color bars (100% bars). Encoded color -- NTSC The RGB system, with parallel full-bandwidth video in three channels, makes excellent pictures but has several practical disadvantages. Perhaps the most severe is the very wide bandwidth needed in broadcast service or over other long interconnect paths. There have been several attempts at "compressing" color video information so it is more practical to record, interconnect, and broadcast. The winner, by far, is the NTSC system developed in the early 1950s. This "encoding" scheme manages to put the essential parts of the three-dimensional color signal into the spectrum needed for a monochrome signal -- putting it all in one path. (Work is still continuing today with complex digital filters, advanced modulation techniques and "smart" compression algorithms, but these are beyond the scope of this tutorial.) NTSC must still maintain three information paths, inherent in the nature of color. The specific set of three signals, however, is changed from RGB to a more manageable set. The three NTSC signals are Luminance, and two Color Difference components. The symbol used for luminance is "Y." It is closely related to movement in the "y" direction in a common x vs y graphical representation of color. The color difference signals also contain Y and one of the primary colors -- usually B-Y and R-Y. For example, B-Y is the Blue video signal minus the luminance signal. Luminance The NTSC luminance component is very much like the signal for monochrome video -- containing information about the amount of light in each element of the picture. This is one of the key reasons the NTSC system is "compatible" with older monochrome equipment -- monochrome sets simply ignore the color information and display using luminance only. Luminance is derived from RGB color signals. While the tristimulus theory requires three different colored lights, it does not imply that each light contributes the same amount to our perception of brightness of a point in the image. In fact, for the particular colors chosen (a certain Green, a certain Blue, and a certain Red), Green contributes about 59% of our perception of brightness in a white part of the image. The Red light contributes about 30%, and Blue only about 11% of the brightness of white. These numbers are called the luminance coefficients of the primary colors. (A certain color of white must be chosen as well, but that and the math involved are beyond our scope here.) The luminance signal is produced by combining (adding) the video signals from each primary color channel, taking into account the luminance coefficients of each primary, i.e., Y = 0.59G + 0.30R +

0.11B. If the input signal for this process is a RGB color bars signal, the resulting luminance video waveform is the familiar decreasing "staircase" shown in Figure 7-2.

Figure 7-2. Color bars luminance signal (100% bars). Note the steps in this staircase are not equal amplitude -- they depend on the luminance coefficients. Consider, for example, the Green to Magenta transition (in the center of the bars waveform). The luminance level for Green is 59%. Magenta consists of Blue and Red only, its luminance value is 0.11 + 0.30 = 0.41 or 41%. The amplitudes of the other levels may be similarly calculated. Color difference signals -- B-Y and R-Y Once the luminance signal is available, it takes only a little more processing to derive the color difference signals. For example, if luminance is subtracted from the B component of the RGB video the result is B-Y. This process, and the parallel one that produces R-Y, are fairly easy to accomplish with electronic circuits. (Sometimes, another set of color difference signals were used -- I and Q. We'll discuss I and Q a little later.) With the color information now in the form of luminance (Y) and color difference components (B-Y and R-Y) we still have three signals and the associated bandwidth and interconnect problems. A bit more processing is needed to get it all into the "'one-wire" NTSC signal format. Encoded color (NTSC) Another characteristic of human vision is we can't see fine detail nearly as well for changes in coloring as we can for changes in luminance. In other words, the picture won't suffer very much if we reduce the bandwidth of the coloring components, provided we can maintain essentially full bandwidth of the luminance signal. In fact, this is a good reason for developing color difference components in the first place. Even a full bandwidth luminance signal doesn't have very much energy in the upper end of its spectrum -- the higher frequency signals are quite a bit lower amplitude almost all the time. These two facts (less bandwidth required for the color information and some "room" available in the luminance spectrum) allow the NTSC system to place the color components in only the upper portion of the luminance spectrum. Color subcarriers The technique for getting the R-Y and B-Y signals moved up in frequency is amplitude modulation -- a carrier (subcarrier) is modulated by each of the color difference signals. Remember, there are two different signals, so two carriers are needed. Both signals are at the same frequency (about 3.58 MHz)

but are maintained in phase quadrature (90 degrees apart) so the B-Y and R-Y information can be kept separated. This process is known as Quadrature Amplitude Modulation (QAM). Balanced modulators are also used so the carriers themselves are suppressed at the modulator output. Only the sideband energy generated in the modulation process is saved. This modulated color difference information is usually called chrominance. The NTSC composite signal When luminance and chrominance are combined, the result is the NTSC composite color signal. Figure 7-3 shows the B-Y and R-Y signals for color bars and the resultant envelope at the output of each modulator, as well as the combined chrominance envelope containing both B-Y and R-Y information.

Figure 7-3. Unmodulated and modulated color difference signals and modulated chrominance for color bars. Some adjustment of amplitudes is needed when the luminance and chrominance signals are combined. Figure 7-4 shows the result if the signals for full-amplitude bars are combined without these adjustments. Note the peak level of the chrominance is very much above 100% white luminance. This signal will be seriously distorted in passing through a video system, especially a system originally designed for handling monochrome signals. In Figure 7-5, the level of B-Y is reduced by a factor of 2.03 and R-Y level is reduced by a factor of 1.14. These specific factors are chosen to limit the peak of the waveform to 100% white level with a 75% amplitude bars signal (Figure 7-6). This 75% bars signal has become the most common for testing NTSC systems. (Refer to A Closer Look at the Color Bars Waveform and Beyond Bars - Other Tests and Test Signals for the discussion of various bars signals .)

Figure 7-4. Full amplitude bars without gain adjustments.

Figure 7-5. Full amplitude bars.

Figure 7-6. Corrected 75% bars signal (with 100% white). Conclusion There are several aspects of the generation of NTSC video signals that underlie recommended monitoring and test methods. 1. Luminance and Chrominance are a "matched set" and must remain balanced in amplitude and timing in order to accurately reconstruct the RGB signal (and the color picture). 2. The color difference components are only kept separate by differences in subcarrier phase. Small phase distortions can create large color distortions in the end picture. 3. Chrominance information is all at the upper end of the video spectrum and maintaining frequency response flatness throughout the system is critical. The color bars signal, whether in RGB or NTSC form, offers a known, accurate signal that exercises the practical limits of the color video system.