A Study in Contrasts

Contrast: The ratio between the brightest part of the picture and the darkest.

You might not find that precise definition in your Webster's, but it's the one that applies to our little corner of the world. It's also one of the most important characteristics of a video image.

But there are a number of ways to measure contrast—more precisely contrast ratio—and that's where the problems begin. One method is so-called ANSI contrast, which measures the contrast between peak white and total video black when both are present on the screen at the same time, usually from a standard checkerboard pattern of black and white squares. The presence of white and black together is the toughest test for contrast ratio, since the white areas inevitably compromise the black.

But ANSI contrast comes with its own baggage. Since room reflections can seriously compromise the result (by reflecting light back onto the black areas) it is most accurately measured in a theoretical "black hole." Such a totally non-reflective environment does not exist (the audio equivalent would be a perfect anechoic chamber). And the technology with the deepest blacks to date, both real and subjective—CRT—produces unimpressive ANSI contrast numbers.

The other method, peak contrast, is the one we feature here at Ultimate AV. It's also the one used most commonly by manufacturers in their spec sheets. How manufacturers arrive at these numbers is another story entirely, since our results rarely verify the contrast ratio specs. We suspect that the factory numbers are often derived at picture settings you wouldn't use for normal viewing. Our numbers are made only after we have arrived at the optimum settings—the settings that produce the best picture.

But peak contrast is only meaningful if it includes the two numbers it's derived from: the white level on the screen when the signal is a 100 IRE test pattern (100 IRE is the theoretical peak brightness present in a video source), and the level of total black when the image is a full screen, video black test pattern. If the peak white level is high enough you can have a high contrast ratio even with sucky blacks. In real-world viewing, that will virtually guarantee a bad picture.

Both of these figures can be measured with reasonable accuracy with a high quality meter (we use Konica-Minolta LS-100, a $3000+ device that is more or less an industry standard). We measure the light from the screen in foot-Lamberts (fL), and also specify the size of the screen and its gain when we're discussing a video projector.

In measuring peak contrast, the room reflections are not particularly significant as long as the room is totally dark to begin with. When measuring the peak white level, the image on-screen is so bright that room reflections are inconsequential to the reading. When measuring the video black level, the light output is too low to produce significant room reflections back onto the screen.

But the black levels of a few projectors, in particular, are now getting so low that they are scraping the lower limit of our light meter's specified capability (0.001fL). So what to do?

In my recent review of the JVC DLA-HD100 projector, I fell back on a technique I have used in the past to verify the readings I normally take on my full, 16:9, 78" (wide) screen. If you project the test pattern to be measured on a much smaller screen, both the peak white and the video black will be raised to a level well within the comfort range of the meter. Since they should both be raised by the same amount, the contrast ratio should be an accurate reflection of the true peak contrast ratio you would get off the bigger screen if you had a light meter with a far more extended low-level range (a meter that only a high-profit conglomerate or government agency could justify!).

When making this "small screen" measurement, nothing in the setup is changed. I do not move the projector or alter its settings, nor do I change the zoom setting of the lens. Instead, I merely position a sample of the same screen material (Stewart Studiotek 130 in my case) close to the projector. The only compromise is that the light meter is positioned a few degrees off to the side to keep it at the height of the lens and still have a clear shot at the "screen" without the body of the projector getting in the way.

This method gives us an accurate peak contrast ratio even from a projector with state-of-the-art blacks. But taken by itself it will not provide black level the projector will produce from the full, main screen. Measuring the black level of the JVC on the main screen produced a black level of 0.001fL. That's as low as the meter will read. It does tell us that the blacks are really, really deep. But even ignoring possible meter errors at this low level (and there are meter tolerances in any measurement) the actual figure could actually be as high as 0.0014fL, since the meter rounds off to three decimal places.

At this point I resorted to a little mathematical sleight of hand. I took the peak contrast ratio obtained with the "small screen" method described above (16,000:1 in the case of the JVC), and divided it into the peak white level that the JVC produced on the full-size screen (about 13.5fL) to determine the effective black level we'll see on the full screen. For the JVC DLA-HD100, the result was 0.0008fL.

While direct measurements are always best, interpolation such as this is not uncommon when arriving at a result by direct means is either not possible or beyond the reliable capability of the measurement tools. And it can provide useful comparisons. For example, when I measured the black level of the JVC-DLA-HD100 from the big screen, I obtained consistent readings of 0.001fL. Doing the same for the JVC DLA-RS1 resulted in readings that alternated between 0.001 and 0.002. This made a meaningful numerical comparison of the two projectors difficult. All I could reasonably conclude is that the blacks from the HD100 were a little darker. But when I applied the same small screen/large screen technique to the RS1, I obtained a black level of 0.0014fL, telling me that the black level of the HD100 really is close to half that of the RS1—a result that could not easily be obtained directly.

It's unfortunate that this technique can only be used for projectors, not for one-piece HDTVs. But in the latter case a few manufacturers are approaching the point where their sets will literally shut off their drive circuits when presented with a video black signal. And no known meter can take a meaningful reading in that situation!