Machine Over Man: Coloring Grading Technology

By Ronit Scheyer

That the perception of Color is subjective is no secret. Scientists and medical experts have solidly traced color perception in terms of biology to light wavelength receptors within the retina of the eye, processed by the part of the brain’s cortex called the ventral stream. But how humans describe, define and distinguish shades and hues from other shades in the spectrum – and whether one human will see a certain color the same as another – is an entirely different matter.

Anyone who knows even the basics of diamond grading knows that a final grade for a diamond given by a laboratory is based on four main factors, popularly known as the 4Cs – carat weight, clarity, cut and color. It is with this last factor that the subjective human element of color perception is involved.

The fact that color perception is subjective means that no expense has been spared in removing any element in the color grading process that would cause the grader to give an inaccurate grade.

In most labs, color grading is performed by trained, expert gemologists, who examine individual stones in comparison to a set of carefully selected and regulated master stones, in highly controlled conditions.

Regulations in these labs have been established, preventing color graders from working past 1 or 2 in the afternoon because perception of color changes as the day progresses. Conditions in rooms used for color grading are closely regulated to prevent outside light, wall color and even the color of one’s clothing from influencing what the eyes would see when a grader set down to grade. Color graders are even required to sit in the room for at least 10 to 15 minutes before grading a stone, to let their eyes adjust to the light in the grading room.

For all these reasons, it seems that there is room for technology in the grading process.

Technology is not a new phenomenon in color grading. Belgian lab HRD Antwerp NV uses a color grading instrument called the Lumisense, which, the lab says, “provides a link between human color observation and instrumental color measurement.” However, the lab declined to provide details on the machine’s process, what kind of technology it utilizes or what, exactly, it measures.

Sarin Technologies Ltd has developed a color grading instrument – the Colibri – that acts, in a laboratory environment, as a “second opinion” for color graders. The Colibri, which replaces Sarin’s previous model, the DC 3000, was officially launched at the end of 2007 and evaluates the color of a loose or mounted round or fancy shaped white diamond (D-Z) according to the color grading parameters of five major labs – AGS, EGL (international), GIA, HRD and IGI.

The Colibri is calibrated according to the “main” master sets of these five labs, meaning that it does not grade stones independently. What it does mean is that a diamond lab (or retailer or manufacturer) can, for example, insert a stone into the machine and find out that a certain lab’s color grading parameters would give the certain stone an F grade and a different lab would give the same stone an E or E- grade, something that has major implications for pricing and to which lab a manufacturer might send their stone. If a company wishes to grade independently, they can upload their own standards, or master set, into the machine so that it will, in the future, grade diamonds according to that master set’s color parameters.

The machine also detects fluorescence – its color and its strength – can print a label when hooked up to a printer and can upload the information, plus other 4C’s information entered by the user, through a USB port to an Excel spreadsheet. It can also run on battery power and is relatively light, making it mobile.

Relying solely on technology to grade a diamond’s color is not at all widespread. In fact, the vast majority of labs, according to Sarin’s Vice President of Sales and Customer Care Natan Mossak, do not use any technology whatsoever in the color grading process. Most diamond grading labs still grade for color the traditional way – with the human eye.

International Diamond Laboratories (IDL), a subsidiary of the Dubai Multi Commodities Centre, is one lab exception. The infant lab, based in Dubai and founded just a short while ago, grades white diamonds (D-Z colors) with a machine called the IDL Color Meter™. The technology works by measuring the spectrum of the diamond – the amount of light, of the light that falls on the stone, absorbed by the stone.

The process is quite complex and utilizes “a mathematical method called the trisimulus method, [which] assigns a coordinate in the 1931 CIE (International Illumination Committee) diagram, associated to the measured spectrum. This can be converted into a saturation value and dominant wavelength,” giving the stone a color definition, which, in turn is translated into a grade according to the GIA grading system of D-Z.

What this means, according to the lab, is that the process “is not based upon the personal perception of a grader or a non-regulated set of master stones, but [on] an automated process,” which, says Chief Quality Officer Peter Borgmans, succeeds in completely removing the human element from the color grading process.

The importance of the IDL Color Meter, says Borgmans, is that it is a scientific approach to something that has been until now subjective. There are limitations in visual observation, he explains, which means that the results are not always consistent. Many elements must be taken into consideration in the color grading process, including the viewing angle, the distance of the stone both from the eye and the light source, the light source itself, the surrounding environment and, most importantly, the human eye itself.

The human eye, he says, does not have the same capability for making small distinctions between different amounts of color. Because of this, and because, with humans, there is something really personal, including mood, that can affect grading, the lab looked for something that would eliminate the possibility of an inaccurate grade.

 In the process of calibrating the machinery to certain color grade parameters, IDL performed analyses of thousands of stones, with certificates from various labs, and their color grades. The lab then measured the amount of color in the stones and plotted their grades, coming out with an average value for each letter grade, D through Z. So, in reality, as Borgmans explains, where human eye grading is completed with a single master set of diamonds (one stone for each letter grade against which the stones being graded are compared), the “Color Meter grades according to a methodology that is unique, that has been proven to be fully consistent after being scientifically tested and that is not based anymore on a set of master stones.”

“The major advantage of this is that we can reproduce our measuring results over and over again, and the grading is consistent. We also performed research on human eye grading, and we saw an error margin of more than 25 percent. This means more than 25 percent of stones graded by the human eye in our test had an inconsistency,” Borgmans explains.

 “Honestly, I don’t see any drawbacks in our system. We perform regular checks with stones. Our scientists are very closely monitoring the system and measuring it with a large number of stones again and again, running internal quality checks, and I am very confident.”

However, not everyone is so confident that it is possible to do what IDL says it has achieved – completely remove the human element, and therefore the possibility for inaccuracies. Menahem Sevdermish, founder of the gem software company Gemewizard and a well-known authority in the field of gemology, for one, strongly asserts that the integration of such technology into color grading is impossible. “Color grading is subjective, no matter which way you look at it, no matter which way you turn it. There is no way to remove the human element from the grading process,” he says.

According to Sevdermish, even the machines are calibrated according to master stones that have been given certain grades by humans. “Machines represent and grade according to the thousands of humanly graded diamonds which were used to calibrate them.” Two human beings who are trained, expert gemologists may give a single stone the same color grade, but no two human beings will see the color the same. Furthermore, no two diamonds are the same, so to calibrate a machine according to what humans have graded means that the possibility for human error is integrated into the machine.

 Borgmans concedes this point, saying, “It is indeed impossible to create a perfect set of master stones. This is also based on visual observation (human eye grading).” But he stresses that, exactly because of this, they do not use a set of master stones. “We are doing it another way, we are avoiding this problem by setting objective criteria and limits that are measurable and that are derived from the huge statistical analyses. The high consistency removes the problems related to the subjective human interpretation, i.e. a stone getting an E grade yesterday getting a D today by the same grader – it is this human element that must be removed.”

An additional problem is that the presence of color in most diamonds is not uniform throughout the stone. It may be darker in some areas than in others, and its appearance may change when the stone is turned certain ways or looked at by different people at different times from different angles. A stone may be inserted into a machine at a certain angle and inserted again at a different angle, and the result may be a different grade, according to Sevdermish.

“Machines can measure color accurately,” he says, “but they do not see how it really looks. Machines can only complement the work of a highly trained grader,” not replace it, in essence.

Sevdermish’s controversy comes when a diamond’s color falls somewhere in between two grades. “When everyone agrees that a diamond is a G, for example, and the color falls in the center of the G grading spectrum, there is no problem,” Sevdermish explains.

The problem occurs when a color is between two sectors, and the real question is who, or what, decides where that border is and on which side of it a stone falls, a decision that will affect its overall grade and its price. “What happens,” wonders Sevdermish, “if,” hypothetically, “a thousand graders give a stone a G grade and a machine gives the same stone an F grade? Who is right? No one is right, I think.”

"It is all subjective, and if you must decide who is right, then the humans are right, because humans created color grading, and color grades are created for humans.”

But is there ever a possibility that the machinery will make a mistake or give an inaccurate grade? “Per definition, I never say ‘never’,” Borgmans says, after a pause. “But I have to say that I am feeling extremely comfortable.”