Yoshi Ohno, chercheur au NIST (National Institute of Standards & Technology) a accepté d'être interviewé par Laurent Massol, gérant de la société LED Engineering Development (Toulouse, France) et Benjamin Monteil (propriétaire de LED-fr.net). Ceci est une interview exclusive, et nous remercions chaleureusement Yoshi Ohno pour sa disponibilité. La traduction en Français de l'interview sera diffusée sous quelques semaines.
Laurent Massol. The CQS affects the score of lamps which have CCT below 3000K. You give us an example at 2800K that decrease the score by 2points. Could you give us an idea about this new score for a lamp at 2500K? What is your feeling about this (low) reduction, because from customer point of view, between 2500 and 3000K, the lamp is not considered as a "ideal spectrum" but as warm white. So do you think that CQS can in a second step, decrease again the score for such lamps considered by end users like warm and yellow?
Yoshi Ohno. The CQS score for a source with 2500 K CCT would be reduced by about 4 points. The intent of the CCT factor is to lower scores for decreased object color gamut/discriminability. It does not take CCT preferences into account, which is what I think you are describing.
LM. Could you give us an example of spectrum which reachs the score of 100 with your CQS model?
YO. If you have blackbody spectrum from 3000 K to 5000 K or a daylight illuminant at 5000 K or higher (these are reference illuminants for CQS), you will get CQS score of 100. This is the same as the CRI because the CQS uses the same reference illuminants as the CRI.
BM. You said (and you showed) that the old technologies are not affected too much by CQS, and in the same time, CQS gives more information and accurate information on the quality of LED's lamp. So why is it so difficult to adopt it?
YO. The CQS makes the largest differences when sources increase the saturation of object colors. This is not common for traditional light sources (except the neodymium lamp) and that is why CQS scores do not change much from CRI for traditional sources. On the other hand, the increase of color saturation can easily occur with SSL sources (e.g., RGB type LEDs or narrowband phosphors). Increases of color saturation render illuminated scenes more vivid and are often judged to be of higher quality (as long as the saturation is not excessive) and is perceived to be better than decreases of color saturation of the same calculated color differences. The CQS is formulated so that it does not penalize such lights (saturation-enhancing) as much as the CRI does. So, the CQS takes into account the direction of color shifts (direction to increase or decrease color saturation), which makes the score correlate better with visual perception. This is a different approach than used by the CRI. The CRI measures the basic concept of color fidelity, where the score is reduced equally for object color shifts of any direction. Many people who have been using CRI over many years, especially from the traditional lamp industry, have difficulty accepting the CQS's new approach and think that the standard color rendering metric should be on the simple principle of color fidelity. We are concerned that a new metric based on simple color fidelity would not be meaningfully different from CRI and would not solve the problems raised by the SSL industry.
BM.Few days after your paper, the LRC (Rensselaer Polytechnic Institute) proposed a new index, which is named "GAI". What are the main differences between CQS and GAI from you point of view?
YO. The GAI is to supplement the CRI, thus not designed to replace the CRI. The GAI would have very limited usefulness by itself, as a high value of GAI does not necessarily mean the color quality is good. Actually excessively high GAI scores would not be good. To make it work, both numbers (CRI and GAI) must be shown for one product. The interpretation of these two numbers would need expert knowledge, in my opinion.
The SSL industry (and the rest of the lighting industry, for that matter) likes to have a simple one number for color rendering rating (like the current CRI works). To meet such needs, and solve the problems of the CRI, the CQS is designed to provide an independent one number metric, intended to replace the CRI. We believe a simple one-number metric (to rate the overall color quality of the product) is needed for general consumers, without requiring expert knowledge, and additional specific indices (color fidelity, color preference, etc.) will also be useful for expert users.
BM. The industry doesn't accept any new index in order to replace CRI. The need in public lighting are not the same than the need in retail lighting, so the need in a new color index are not the same. From your opinion, is the industry going to "specialized" index (one index per application) ?.
YO. I'm not totally sure that I understand what you are asking. Please let me know if this doesn't answer your question. I don't think specialized indices are a good idea. Multiple metrics would lead to confusion and wouldn't allow lighting products to be directly compared to each other. Different applications do indeed have different color quality needs, but this can be addressed through performance standards (e.g., one could require CQS of 80 for residential spaces, CQS of 60 for roadway lighting, etc.). But, we think that the color quality of all light sources should be measured the same way. I think CQS would be useful for most of general lighting applications. For special applications, additional specific indices as mentioned above could be used, in my opinion.
About Yoshi Ohno
Yoshi Ohno is Group Leader on the Optical Technology Division of the NIST (National Institute of Standards & Technology)
Sources
Official page of Yoshi Ohno on the NIST website
Publication of Yoshi Ohno about CQS, june 2010
Publication of the RPI (Rensselaer Polytechnic Institute) about GAI and CRI