Monday, November 16, 2009

Color Creeks Ray Winey Has Found People Up

Bill Longley, Adjunct Professor, Eastern Michigan University
Retired, Ford Motor Co.

[Who was the first to perform colorant formulation on a digital computer? Bill Longley here makes the case that---at least in the U.S.---it was his mentor, Ray K. Winey. Winey is also known for the metameric specimens he made, which screamed, “Matching under two lights is not enough!” It is hard to make metamers, and also to convince people to avoid mistakes in color science, yet Winey did both. - MHB]

I had the rare good luck after graduating from college in 1958 to work under direction of Ray K. Winey at US Rubber Co (later Uniroyal) in coloring of Naugahyde upholstery material. Ray had majored in chemistry at Notre Dame and spent his career in Mishawaka, Indiana, never wanting to wander very far. When he spotted an interesting conference he would send one of us and then eagerly debrief us upon our return. Ray, now deceased, was a genius with an intriguing grasp of poetry as well as battlefield war strategy. His recall and passion for all aspects of color science was overwhelming to color chemists who came under his direction.

Davidson & Hemmendinger had introduced the COMIC, an analog computer with wavelength/reflectance scope which aided the colorist in developing initial colorant formulations. I persuaded Ray to make the short drive to Chicago to see COMIC at an equipment show. Here we met Henry Hemmendinger, who patiently demonstrated COMIC for the umpteenth time that day. Ray folded his arms and commented, “Very interesting, but I prefer the digital approach.” Henry wanted more discussion, so he called for assistance with the exhibit and then he and Ray moved to a nearby table and launched into an unforgettable session.

Ray summarized his work on digital computer matching in a company report [1] dated 15 August 1962, complete with color standards and resulting swatch matches. The report discusses limitations of analog computing, especially assumption of zero scattering. Rather than requiring the colorist to select pigments for the match, Ray directed the program to consider all possibilities of 3 pigments plus white, and then report the matches with least spectral differences. If dissatisfied with results, he could direct the program to add a fourth pigment. Ray’s report cites Kubelka-Munk and computes relative K and S coefficients, 10 parts TiO2 to one part color pigment. He relates reflection to K/S ratio using the Saunderson surface correction. Computing was slow so he loaded data into the Bendix G20 at close of office hours and collected results the next morning. Contemporaries in the field who have seen a copy of the report have marveled at digitally calculated two-constant theory at that early date. It wasn’t until five years later that D&H offered the COMIC II digital computation unit.

Details and comments on the report are offered by another Winey protégé, Ron Penrod, at Site viewers will also find there an excellent color matching program, workable in four modes with conversions. [Editor’s note: Do any readers know if E. Atherton’s program from the UK (reported in 1961 J. Soc. Dyers Col., and reputed to have run since 1956) had the capabilities described here?]

Ralph Stanziola and Max Saltzman were among many who visited Ray in Mishawaka. He enjoyed showing visitors his “dilemma” samples (two metameric swatches), asking how the visitor would add a correction to the metameric mismatch that was redder than the standard in daylight and greener in tungsten. Told this was impossible, he then presented another swatch containing pigments that flared green to the standard in daylight and red in tungsten. In fact this is how the standard had been made.

Ray had some classic correspondence with Norman Macbeth and Warren Reese concerning Macbeth claims for their industrial light-source unit, that samples matching in D7500 daylight (blue end) and D2300 tungsten (yellow end) would match anywhere. Ray produced numerous samples to disprove the claim. Eventually Macbeth added cool white fluorescent as a third source for matching, also standardizing on D6500 daylight. I like to think that Ray provided the impetus.

I offer these notes here to credit Ray for his pioneering work in color science. I inherited Ray’s “golden” files and sometimes wonder what to do with the intriguing letters and metameric specimens. I have his hand-written note saying he wanted to write a book Color Creeks I Have Found People Up. He never wrote the book, but certainly had enough material for it.

1. D. F. Larimore and R. K. Winey, "Color Matching with the Aid of a Digital Computer," Report No. 62-L3-35, US Rubber Co., Divisional Laboratories, Consumer and Industrial Products Division, Mishawaka, Indiana, 15 August 1962.


Danny said...

I must disagree with my old friend Bill Longley on this article from the ISCC Newsletter. In Eugene Allen's first paper, "Basic Equations Used in Computer Color Matching", JOSA, v26, n9, pp1256-1259,(1966), he cites two earlier papers which he says described as, "Iterative methods for use with digital computers have been described, but the descriptions given are incomplete because the starting values are not specified." Those two articles are:

Billmeyer, Beasley and Sheldon, JOSA, v50, p70ff, (1960) and Alderson, ATherton and Derbyshire, JSDC, v77, p657ff, (1961).

Both of these papers pre-date the report by Winey. Allen goes on to state that Davidson, Hemmendinger and Landry reported on the equations used in the COMIC in their paper in JSDC, v79, p577ff, (1963).

That said, I would argue that the original COMIC was not a digital computer. The COMIC II was the first commercial digital computer for color matching. The COMIC utilized boxes of potentiometers or adjustable resistors. Each primary dyeing was intput into a bank of 16 potentiometers (pots) and the resistance was adjusted to produce an output voltage proportional to the reflectance factor. An analog circuit would take that voltage and convert it to absorbance and then the absorbance of a set of 3 primary dyes was summed using analog electric circuits until the combined voltage matched the voltage of a standard curve. The banks of pots contained 16 individual adjustable resistors, or a sample from every 20nm from 400nm to 700nm.

The Billmeyer report described his work on matching transparent plastics, in which the General Electric Hardy spectrophotomter was interfaced directly to an IBM card punch machine. As the drive scanned the prism monochromator, the photometric encoder made contact with the wires on the cardpunch and a column holes were punched representing the reflectance values in binary coded decimal. When the spectral scan was complete, the card was taken to a card reader and the spectral reflectance reading was read into an IBM 1108 16bit mainframe computer that the Dupont R&D center owned. The matching calcuations would churn and iterate for a while and then spit out a formula. They also had a correction algorithm that could update a first guess. Fred used to show a slide from the paper, or perhaps more likely from a report to his superiors at Dupont, which contrasted the iterations of the matching engine to the iterations of the visual shader.

IsiahGoods said...
This comment has been removed by a blog administrator.