Thursday, August 4, 2016

From Color Science to Privacy

[Here is the latest word from Michael Stokes, architect of sRGB, a founder of ICC, Interest Group II chair of ISCC, …, and ten years absent from the field of color.]

I grew up in my family’s slide duplication and photofinishing business back when E4 processing was still new, and Kodachrome, FujiChrome and AgfaChrome were widely available. The business grew to the point my parents bought entire emulsion runs of motion picture and inter-negative stock from Kodak to ensure consistent quality. Our computer system tracked every action taken of each slide or component of each order, and we offered an unconditional money-back guarantee. I was responsible for the quality and production aspects of the business, so I spent what seemed like endless time trying to understand why it was so difficult to craft color reproduction processes that were consistent and accurate. 

In the late 1980s, I joined RIT’s Color Science Master’s program and soon my frustration transformed in wonder and awe that we were able to reproduce color reasonably well given the many complex aspects involved. I was fortunate enough to have a successful color career at Apple, then at Hewlett-Packard, and finally at Microsoft as their Color Architect. Over this period, I helped to lead industry efforts such as the founding of the ICC, standardization of sRGB and scRGB, evolution of ColorSync and ICM, as well as taking part in the founding of CIE’s Imaging Science Division. I often explained the goal of my efforts as trying to reverse engineer a large part of the human brain to effectively model it across a complex system of components from many stakeholders who didn’t always get along with each other. 

I joined Microsoft in 1999 in hopes of implementing the best-in-class color reproduction software development systems. By 2006 we had successfully implemented advanced hybrid solutions that supported sRGB, scRGB, and ICCv4 as well as CIECAM color vision modeling. This was a part of Windows Vista, which was known to have significant challenges. 

It soon became clear that advanced color systems were not the most pressing need for the company, and I transferred to Microsoft’s newly established Health Solutions Group (HSG). As part of this transition, the company required that I no longer participate in the color field and have no communications regarding color with my many friends and colleagues that I had established over many years. It was a difficult personal decision, but in the end, I chose to put my family’s financial stability first. I am sincerely and deeply sorry for the negative impact this has had on my communications and relationships with many dear color friends and colleagues.

My original job description focused on health standards and patents, in which I had significant experience from my color science background having worked with IEC, CIE and ISO as well as accumulating over 50 patents. Within two weeks, my role grew to include a focus on security and privacy. I had security experience having designed and written color and imaging components in the Windows kernel software. Privacy was completely new. 

It turns out that privacy is very much like color reproduction. The goal is to understand human desires and perceptions around information data flows, controls, and ownership to effectively model these desires and perceptions across a complex system of components from many stakeholders who don’t always get along with each other, including regulators, legislators, and consumer advocates. 

I was again extremely fortunate to enjoy a second successful career, including testifying before the US Senate Judiciary Committee, visiting numerous national and international regulators including the US Food and Drug Administration, US Federal Trade Commission, and CNIL  (French Data Protection Authority).
When Microsoft divested HSG into an independent subsidiary jointly owned by General Electric (and now wholly owned by General Electric), I again chose to put my family’s financial stability first and stay at Microsoft. I now help manage privacy for many of Microsoft Office client applications including Word, Excel, PowerPoint and OneNote among others. I believe I have one of the best jobs in the company, helping design architectures and processes to empower our users to best achieve their desires. 

The major regret I have had in my career was not being able to explain why I “went dark” from all of my color friends. I hope this article sheds some light on those difficult days and deeply appreciate the opportunity that ISCC has provided me. I know I owe two successful careers to my friends and colleagues in color science who taught me so much about color, and even more about life.

Thank you,
Michael Stokes,

Tuesday, May 3, 2016

A Gentle Color Paradox

In my recent web-browsing, I ran across an exercise for students [1]: “If the wavelength of the green line of mercury is 546 nm in a vacuum, what is it in water? In heavy flint glass? [410 nm, 331 nm].”

This awakened me to a question I have not seen posed by color science. Would the color of a light change as drastically as suggested by the wavelengths in the above paragraph? [Strictly speaking, of course, the waves have no intrinsic color, but I use loose terminology for convenience, with the understanding that the only meaningful connection is a cetera paribus color match.]

The effect of wavelength shift would depend upon the context. Surely placing a piece of flint glass in air between a mercury-line lamp and a human viewer would not influence the wavelength of light arriving at the eye. Although the wavelength 546 nm would change to 331 nm while the light was within the glass, it would change right back to 546 nm as soon as it re-entered the air. Then it would change to its value of 410 nm upon entering the aqueous medium of the eye and photoreceptor system. (Of course, the light’s frequency would always be the same.) But we wouldn’t notice any change due to the 546-to-410 nm transition because we always see through the same aqueous medium (with the same refractive index) embedding the photoreceptors. 546 nm as measured by a spectroradiometer is hence calibrated to be identical to 410 nm as incident on our photoreceptors.

But suppose we change the refractive index of the material abutting the eye’s photoreceptors and see if the color of the refracted light changes. Even though abutting the photoreceptors with a non-waterlike substance seems impossible with real eyes, it is an option with cameras whose photosensors are embedded in the chosen refractory material; also, visual prostheses in the future could have this option. 

So, under these circumstances, would the color of a light change as drastically as suggested by the wavelengths recited in [1]? If the color did not change, then for this weird situation one should compute tristimulus values with frequency rather than wavelength as the integration variable: light does not change frequency when passing through a transparent medium. Perhaps someone has already tested the idea with cameras. I await the impact of the idea once prosthetic eyes are abundant.

[1]. HQ Fuller, RM. Fuller, and RG Fuller, Physics Including Human Applications. Harper and Row, 1978. Revised electronic version copyright 2009 by RG Fuller. exercise 2, p. 436.

Michael H. Brill