Projection displays are getting better and better these days, no doubt about it. And to tout these improvements, display manufacturers are quoting specifications for their latest consumer products that include "on/off contrast ratios" in excess of 30,000:1. These numbers are intended to show that the projectors are not only bright, but capable of reproducing extremely low black levels, a traditional weakness of fixed-pixel displays. In reporting these numbers, Insight Media has heard from readers who argue that these numbers are irrelevant, misleading, and next-to-impossible to measure. Partially valid, sure. But, more importantly, we feel the industry is missing the ball when it comes to characterizing the images these products deliver.

John DiLoreto
Analyst and Editor for
Insight Media

First of all, can you measure these extreme levels of contrast? Yes, you can, says Dr. William Bleha, head of D-ILA development for JVC. In a conversation with him this morning, he described JVC’s testbed in Cypress, CA. In two all-black rooms with virtually zero ambient light, they set up their projectors looking directly into a $3,000 IL-1700 Photometer from International Light Technologies that measures lux over a range of 8 x 10-4 to 8 x 105. That’s 9 orders of magnitude, easily handling the range.

JVC uses DisplayMate software on a PC to generate full-screen images of 100 IRE (maximum white) to 0 IRE (computer-level black). Dr. Bleha invited Insight Media to visit their lab any time we were in the L.A. area to verify the measurements of 30,000:1 on their latest projector, the DLA-RS2.

Other projection displays that have quoted extreme contrast ratios include Samsung SP-1800 (10,000:1), Sony VPl-VW60 (35,000:1), Texas Instruments RPTV prototype (100,000:1), SEOS Zorro projector (500,000:1).

But the bigger issue is the meaning of such a measurement. I’ve never been a fan of the on/off contrast measurement. The argument is that this measurement characterizes the darkest level that the projector can produce. But who watches a scene where there is no image? As soon as you add any features to the image, the light begins to bounce around and the black level rises.

It’s the "intra-image" contrast that is important. Dr. Bleha said that when he sets up projectors for Hollywood post-production rooms, "They
want to see the ANSI checkerboard." This 4 x 3 pattern of alternating white and black rectangles gives a much lower contrast, usually in the range of a few hundred to one - quite a big difference.

According to a researcher at 3M Precision Optics, intra-image contrast is limited by internal reflection in the light engine, most particularly the projection lens.

Furthermore, it turns out that the smaller the feature size, the lower the contrast. The ANSI-IT7 specification includes a seldom-used "Small Area Contrast Ratio" (SACR) measurement of 5-pixel black and white line pairs. Contrast in this range is more typically around 10:1.

But even these three measurements combined don’t tell the whole story of how a projector’s contrast falls off with feature size. In the screen business we used to refer to the Modulation Transfer Function (MTF), that showed just how contrast decreased with feature size.

But you can’t use MTF for fixed-pixel displays, informs Insight Media analyst Matt Brennesholtz.

What would help is a more comprehensive contrast measurement for fixed-pixel displays, that’s embraced by display industry thought leaders, such as Dr. Ray Soneira and Joe Kane, according to Bleha. Such a specification should measure contrast at all feature sizes down to alternating pixels. The results would define a curve, the area under which would be a figure of merit for the information content in the image.

Bleha doesn’t expect that manufacturers would publish or even measure these specifications. But perhaps serious publications, such as Widescreen Review, would include it in their detailed reviews of projection products, he said.

We think that this would be a great benefit to the display industry and to serious consumers. What do you think?

Reply to the author