There have been a number of columns recently on display specifications. January 28, Aldo Cugnini talked about color spaces. On November 8, Ken Werner talked about plasma luminance specs. On September 21, John DiLoretto talked about system contrast. This raised such a storm that Insight Media published a follow-up column on contrast on October 8.
Insight Media Analyst
One thread in these columns is a "specification" that doesn’t specify something useful in terms of making a product purchasing decision. Does 30,000:1 with an adaptive iris in the projector actually mean anything?
The 3LCD consortium has pointed out a problem in the specification of color similar to the contrast issue, or the much older luminance issue. The ANSI standard for luminance measurement, instituted in 1992 has largely supplanted the older measurements of projector lumens such as "center-weighted peak lumens" or worse yet, "center-weighted peak lumens equivalent." Center-weighted peak lumens equivalent could easily be 10 times higher than the ANSI lumens figure, so it was beloved by the marketing departments that used it.
In a projector with a white color wheel segment (i.e. no 3LCD projector), the ANSI lumen test reports the total white luminance, produced by the sum of the lumens produced by the red, green, blue and white segments. When the color gamut of the same projector is reported, it is the color gamut defined by only the red, green and blue color wheel segments. It is true this color gamut can be produced, but only at a low luminance compared to the ANSI luminance of the projector.
In some ways, this is actually a good system for displaying colors found in nature. In natural objects, highly saturated colors normally are dim and, with proper decoding, could be reproduced by a RGBW system. The problem is, most projector users don’t want to display only the colors of natural objects. They want to display PowerPoint presentations with bright reds and deep blues. They want to display cartoons with hyper-colors. They want to show flashing light sabers. For that matter, they want to display transmissive natural objects like red, yellow and green traffic signals (if you consider them natural) that can be both bright and highly saturated at the same time.
When a businessman or consumer buys a projector with a RGBW color wheel, or for that matter a RGBCY BrilliantColor color wheel, he would typically have the ANSI lumens and the total color gamut size in the data sheet. There will be nothing in the data sheet saying how bright the saturated colors will be.
At CES Insight Media met with several people from the 3LCD consortium where they explained this proposed test method to us. The consortium, working with Karl Lang of Lumita, a consultancy on imaging and color science, has developed a simple test method for this color issue. The method involves measuring the luminance of the red, green and blue primary colors and comparing that to the total white luminance. They have submitted today (1/30/08) this test method to the IEC as a modification to the specification IEC 61947-1 "Electronic projection- Measurement and documentation of key performance criteria - Part 1: Fixed resolution projectors." This IEC specification is the current definition of "ANSI Lumens," since ANSI withdrew its version of the specification in 2002. The complete text of the proposed specification, plus a paper by Karl Lang on the science behind the specification, can be downloaded from the Lumita website, www.lumita.com. According to the press release from the 3LCD consortium, "Studies are underway around the world to confirm the utility of this new metric." The Lumita study is the first of these studies to be published and I am looking forward to seeing the others.
I have actually used a metric like this myself in the past, where I found qualitatively and unscientifically that a system has poor perceived colorimetry showing video material if the white segment produces more than about 30% of the total luminance. Lang shows data on a group of 9 different RGBW DLP projectors and each one produces about 60% - 65% of its luminance from the white segment. While I think the proposed test may oversimplify the issue and we can certainly expect to hear a rebuttal from Texas Instruments, clearly some metric is needed so the purchaser of a projector can understand what sort of colorimetry he can expect from his projector.