Spring-cleaning: Calibration clarifies the picture, Part 1

While preparing for last summer's video-compression hands-on project, I thought about which of the various displays at my home office might be appropriate for use in the subjective analyses I'd planned (references 1 and 2). My Princeton Graphic Systems AF3.0HD, a 32-in., 16×9-aspect-ratio—that is, wide-screen—HDTV direct-view CRT monitor was an obvious candidate; it is the preferred display for many video equipment and content reviewers (Figure 1). But I quickly realized that, unless I first calibrated it, I would have no confidence in the results I saw.

Just as the characteristics of a speaker or other transducer can affect your impressions of music you play through it, the luminance and chrominance attributes of a display can significantly alter your impressions of images you view on it. And, just as the acoustics of a listening room can transform the frequency response of a transducer, the ambient lighting of a viewing room can interact with a display's output before the photons reach your eyes (Reference 3). As many of us first learned in high-school-chemistry classes, unless you understand all the relevant environmental factors in an experiment and the only condition that changes is the variable under test, the results are of limited value.

How could I accurately judge the degree to which a video-compression algorithm transforms a test clip's contrast breadth and color gamut, both absolutely and compared with other codecs, unless I first dialed in the display settings to points at which they'd simultaneously maximize the possible contrast and color ranges and minimize the distortions at the spans' extremes? And how could I tell if a codec was altering the images' resolutions and aspect ratios unless I had first accurately preconfigured the display's geometry? I'm no expert in display technology, but, fortunately, I knew where to turn for help.

An e-mail to Joel Silver, founder and president of the Imaging Science Foundation, put me in contact with Robert Busch, founder of Busch Home Theater. Busch traveled from Santa Rosa, CA, to my Sacramento home and spent a half-day teaching me both the broad strokes and the fine points of how to properly calibrate my AF3.0HD and, more generally, any display. The information that follows is a composite of insights I obtained both from Busch and from various published tutorials (references 4 and 5). Note that this article has a CRT slant; CRTs are today's display technology of choice for situations in which absolute black level, wide color gamut, and high color accuracy are valued attributes. Some of the issues I raise, such as blooming (an overdrive of the CRT phosphors), are not applicable to other display technologies, such as LCDs and DLP (digital-light-processing), LCOS (liquid-crystal-on-silicon), and plasma displays.

One term bears definition before you continue: IRE level. IRE (Institute of Radio Engineers) is a means of measuring brightness as a relative percentage of total brightness. A 1V p-p video signal subdivides into 140 IRE units, each 0.00714V in size. Absolute black is 0 IRE, and peak white is 100 IRE, or 714.29 mV; synchronization signals extend below absolute black to –40 IRE. In 1953, NTSC (National Television Standards Committee) specifications redefined absolute black from 0 IRE (used in early black-and-white television) to 7.5 IRE, because early black-and-white transmitters couldn't manage a color signal with a 0V black level. ATSC (Advanced Television Systems Committee) specifications return absolute black to its original 0 IRE definition.

When comparing two sound sources, our ears tend to judge the louder one as better, especially after only a brief audition during which there's insufficient time to perceive nuances of distinction. Similarly, our eyes, within reason, automatically judge a brighter display as better. This sensory interpretation, well-known to television salespeople, is particularly relevant to the topic of this article when you consider the high ambient lighting in electronics showrooms and living rooms, with which the photons leaving the display must contend. The end result is that most manufacturers ship televisions from the factory with an excessively high-contrast—that is, peak-white-level—setting. Sales personnel and consumers often further misadjust this setting.

Several problems crop up when you set the absolute-white level too high. For one thing, you lose the fine shades of differentiation between near-white and absolute-white areas of an image. For example, setting this level too high artificially boosts a 98-IRE region to 100 IRE or beyond. Also, overdriving a CRT display's power supply leaves it unable to maintain a constant anode voltage over the wide variations in beam current that result from dark-to-bright and bright-to-dark changes. Within a single video frame, this power-supply instability manifests itself as image distortion, such as a curve in a supposedly straight line. From frame to frame, such as when a lightning bolt suddenly illuminates a dark night sky, the result may be a noticeable expansion and subsequent contraction—or vice versa—of the picture, or, in extreme cases, protracted frame-dimension oscillations.

Excessive beam current also results in blooming, resulting in illumination of an area of the display that is larger than the source image intended. Space-charge effects in the electron beam cause electrons to repel each other, resulting in a wider beam than you intended. This problem is especially critical with HDTV displays, in which the CRT-phosphor-cluster size is often the critical factor in defining the deliverable image resolution. Finally, and perhaps most critical for anyone who's made a multi-thousand-dollar investment in a high-quality display, this excessive beam current may distort the display's shadow mask and will "burn" the display's phosphors, distorting and limiting their color output and drastically shortening the display's usable life.

How can you properly adjust display contrast? Remember that, although you don't want to set the peak white level too high, you also don't want to set it too low, or you'll end up with dingy off-whites at the upper end of the contrast range. First, turn on the display and let it warm up for at least a half-hour before proceeding with the calibration. A pattern containing both 100-IRE and just-under-100-IRE regions is one valuable aid; dial in the contrast setting until you can just differentiate between the two regions (Figure 2a). Other images can help you identify if you've set the contrast too high. In a needle-pulse pattern, see whether you can identify distortion in the vertical white and black lines; such distortion indicates intraframe instability (figures 2b and c). When toggling back and forth between standard and inverse versions of overscan or other patterns, you might be able to detect interframe expansion and contraction (figures 2d and e). In patterns that combine same-sized regions of different IRE levels, look for a difference in perceived size between high- and low-IRE regions, indicating blooming (Figure 2f). A fine-resolution pattern, such the luma multiburst, is also useful for detecting excessive beam current (Figure 2g).

After you dial in the peak white level, turn your attention to the black level, which the oddly named "brightness setting" controls. Contrast and brightness interact with each other; in fact, Busch adjusted the AF3.0HD's black level before adjusting the white level. So, after you dial in the black level, you should readjust the white level. You might need to repeat this process several times to nail down the best possible compromise between the two gray-scale extremes. Improperly set brightness does not cause image distortion, loss of resolution, or phosphor damage such as what you see with an excessively high contrast setting. However, an incorrect brightness setting excessively compresses or expands the low end of the gray-scale range. If you set the brightness too low, pixels that are supposed to be dark gray turn black, and if you set it too high, black pixels become muddy gray. Setting the best possible black level is particularly important with fixed-pixel displays based on DLP, LCD, and similar technologies, which cannot reproduce absolute black.

PLUGE (picture-lineup-generation-equipment) patterns are useful in setting the correct black level (Figure 3a). On the left side of the image, you see three vertical rectangles—one at –4 IRE, one at 0 IRE, and one at +4 IRE. Adjust brightness until the –4-IRE rectangle is not visible, the 0-IRE rectangle just disappears from view, and the +4-IRE rectangle is still visible. Most displays do not hold a constant black level across images with varying APL (average-picture-level) IRE, so, if possible, you should examine the black level across 0, 25, and 50% APL PLUGE levels. The 25% level has a 50-IRE region on the right half of the image, and the 50% level has a 100-IRE background on the right half of the image (figures 3b and c).

Remember that the photons streaming—or, in this case, not streaming—from the display are competing with the ambient lighting of the viewing environment for your eyes' attention. To correctly set black level, you need to observe the display while in a dimly lit room and, to obtain maximum gray-scale and color fidelity from the display, you also need to subsequently operate it under low-ambient-lighting conditions—not in a completely dark room, however, which would quickly result in excessive eye strain (figures 3d and e). Watching television under subdued lighting might at first feel odd, but you'll quickly get used to it, and you'll be pleased with the results (see sidebar "Spousal endorsement and the rest of the story").

Now that you've dialed in the display's gray-scale range, it's time to focus on color and other important attributes. The next issue of EDN will tell the rest of the tale.

Author Information

Technical Editor Brian Dipert is unsure about whether he'll ever again inhabit the cramped, low-fidelity, image-blurry, and popcorn-sticky confines of a movie theater. You can reach him at 1-916-454-5242, bdipert@edn.com, and www.bdipert.com.