Last week, the Society for Information Display’s Mid-Atlantic Chapter had its monthly meeting in the TV testing lab of Consumer Reports (CR), the well-known consumer product-testing magazine headquartered in Yonkers, New York. CR’s Claudio Ciacci and Rich Sulin demonstrated their testing procedures on half a dozen sets distributed around the periphery of the darkened room (including two 3D-TV sets, one plasma and one LCD), and comparisons were inevitable.

Ken Werner
Senior Analyst and Editor

Most of what CR does would not be at all surprising to people with experience in display testing or calibration. Ciacci noted that the testing they do for their own education, which is pretty much what they did for the attendees at the SID-MAC meeting, is more extensive than what they routinely do for the ratings published in the magazine. This is partly because of time and resource constraints, but also because CR’s readers are largely non-technical. Therefore, the testers select those tests and those characteristics that will be most meaningful to the magazine’s audience. Suffice to say that Ciacci and Sulin know what they’re doing, and they can make a sensible case for leaving out what they leave out.

Let’s talk briefly about those inevitable comparisons of the sets under test (without mentioning brand names or model numbers). Subjectively, the best images came from a high-end microdisplay-based front projector, followed closely by a plasma TV (remember, this was in darkened room). Some of the tests that supported the subjective reactions were black level, resistance to motion blur, and color linearity. (More to come on that last one shortly.) It has been a given for years that LCDs have high black levels: in a darkened room, what should be black is actually a surprisingly bright gray. That can have a significant effect on the mood of dark scenes in a movie and reduces the "snap" of high-contrast scenes. Even with local dimming, this remains a significant issue, although Ciacci mentioned he has seen sets recently that turn the backlight completely off when there is no video signal present.

One test pattern I hadn’t seen before — two white circles moving slowly in opposite directions across a black background — was designed to stress a local-dimming system. On an LCD-TV with full-matrix local dimming, the luminance of the circles changed considerably as they crossed zones and the system struggled to keep up. Another set with edge-lighting and vertically defined zones maintained somewhat greater luminance uniformity as the circles moved across the screen.

A standard test for mura — stationary blotchy luminance irregularities across the screen — is to simply put a uniform mid-gray on the screen. At CR, one LCD-TV from a Tier 1 manufacturer had such serious mura it was obvious on the local-dimming test screen. It was the worst I’ve seen in years.

Now for CR’s original approach to comparing color performance. Ciacci showed us the color triangles for a variety of TV sets on a CIE 1931 chromaticity diagram. They all looked pretty much the same, which means panel makers are getting the color coordinates of their RGB primaries right. However, Ciacci said, they noticed that while the triangles looked the same, many sets showed images that looked very different from each other. So Ciacci started playing. One thing he did was plot the 0%, 10%, 20%, etc. color levels on the chromaticity diagram. For red, the ten points should radiate from the white point that is more or less in the middle of the color triangle and go to the red vertex of the triangle in a straight line. For yellow, the ten points should for a line from the white point to the yellow portion of the side of the triangle that connects the red and green vertices, etc. And the points should be equally spaced. That was more or less true for some sets, but not for all. For many sets, points were spaced non-linearly. At worst, points from 60% intensity to 100% would all overlap at the border of the triangle. The lines defined by the points were not necessarily strait, some having alarming curves, which implies errors in color tracking.

To make comparisons easier, Ciacci created smaller color triangles by connecting the 40% points on the red, green, and blue lines defined by his points. As he flipped between the chromaticity diagrams for different sets, you could easily see the size, shape and orientation of some of the 40% triangles vary greatly.

I had never seen this way of presenting chromaticity data before, and it seemed as if the other members of the audience hadn’t either. When it was suggested to Ciacci that he should write a paper on the technique, he seemed to indicate he was too busy to do so. So, I’m presenting this mention of his work here to make sure that credit goes to where it is due, and to allow others to explore the ramifications of this way of presenting TV-set color-performance data.