Color grading of diamonds

[oldminer]

I often think about how many difficult decisions appraisers, dealers and consumers make in regard to color grading diamonds. These decisions have a large economic impact and also are part of the lack of trust felt between conumers and the trade.

Would you like the color of diamonds given in more precise steps than is presently the case with the D-Z GIA system? Crossover points will never be eliminated, but each diamond would be better defined.. Remember GIA color (and clarity) grades are closer together at the top.

Two alternative strategies for categorization could be employed:

1. Numerical grading in even steps. There would be more steps in the darker stones, but the higher quality diamonds would still be more precisely defined. This is the attached image.

2. an even number of numerical steps from color to color. This would finely tune the highest colors way beyond anything a person possibly detect.

Obviously, this sort of work cannot be done without high tech equipment. Would this be viewed as an improvment or just a further impediment to business?

In the long run, I would expect to see a migration to a more refined scale than what we use now. Right now, we have enough problems, and even high tech machines can't yet render 99% error free color grading.

 

[strmrdr]

Id love to see a linear scale model adopted but I would settle for accurate, consitant and repeatable grading within the current system as a huge upgrade.

Id love more steps from a technical standpoint but it may be a nightmare marketing and pricing wise.

Iv thought about this a lot since the last time a simular question was asked and havent hit on what id consider the ideal solution.
There are going to be trade offs in any system.
Where to best put those tradeoffs is the trillion dollar question.

 

[Garry H (Cut Nut)]

Life was never linear.

A small imporvement at the lower end of wine makes little difference to a consumers palate.
A $2 bottle may actually be better wine than a $3 bottle.

But (brand names aside) a very small improvement on a $50 wine can cost $100 or more.

 

[valeria101]

Perhaps the question was intended for sellers, but ... thanks to the open forum I couldn''t refrain from letting the following two points in the open:


#1. Precission and consistency are good things and surely machines can be better than people at that. As long as the reliability of the technology behind this is mentioned to make a ''quality'' argument - why not.

#2. How prefferences are set and buying decissions are made may not change as much due to better reliability, but the relation between the old scale and the new. And that may be settled in the very beginning.


I am a bit puzzled by the current color scale - setting finer steps towards the week saturation seems to go against the fact that the perception of brighteness is logarithmic.

 

[perry]

I vote for precission and repeatability. I am not sure that a new scale would necessarily make it better. Although, a numerical scale would make a lot of sense. However, as Valeria (ana) mentioned, and Cut Nut alluded to, it would have to be a logrithmic scale to make any real sense. Not that logrithmic scales are bad; level of sound (db scale) is logrithmic.

Perry

 

[Capitol Bill]

Dave,
I applaud any effort to bring more precision to any aspect of diamond grading as the technology becomes available. If proven technology exists to grade color, cut or any other characteristic more accurately I''m all for it. Adopting proven technology for the sake of improved accuracy is not a novel concept -- I think it has something to do with a little thing called SCIENCE.

It''s sad to think that a major lab involved in the SCIENCE of gemology (that shall go unmentioned because of intimidating language that hints at pursuing litigation against anyone who disagrees with them) has chosen the argument that there''s no need to pursue more accurate means of measurement because the average human observer can''t discern the difference. In fact, as you are aware, this unmentioned lab admits to intentionally "rounding" its more accurate measurement data because of their "average human observer" argument.

There...I feel better now that I''ve had a chance to rant.

Bill Scherlag

 

[oldminer]

One interesting private message I received questioned the value of finely divided grades with this very good arguement. It is as follows:

Right now, all D color diamonds are "the best" color. If someone adopts a system where D begins at 1 and ends with a number such as 29, then those diamonds at 25 to 29 become instant losers. Every GIA color will have a low end subset that everyone will want to avoid even though they cannot see any difference.

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

I see the point of such a message, but what about the D colors in the 1 to 6 range? Will these not be worth a premium price over other D colors sufficient to make the market back into balance? Do we doubt the subtle strength of a free market when it comes to these stones?

It probably would be most useful to have an even number of steps in each color. I suppose this would be some version of logarithmic. The problem is the unscientific spacing of the existing GIA color grade scale. It is not perectly spaced from color to color, especially in the H I J regions.

All your thoughts are appreciated....

 

[valeria101]

Date: 10/13/2005 10:09:22 AM
Author: oldminer


It probably would be most useful to have an even number of steps in each color. I suppose this would be some version of logarithmic. The problem is the unscientific spacing of the existing GIA color grade scale. It is not perectly spaced from color to color, especially in the H I J regions.

All your thoughts are appreciated....

I do not know more about the technical point about color perception than there is in the study posted on the ADAMAS lab website. There is extensive research about color perception elsewhere (say, related to electronics) so extensive background could be used to refine the current grades. For better or worse, as much as I understand these, the technical arguments support splitting the more colored grades not the colorless ones. Perhaps that economic point has already been exhausted by the current system. In fact, I wonder if the new grid would not be more useful for the 'cause' of light yellows than whites. Also, there seems to be more confusion in the grading system for the near colorless grades that are tinged with something else than yellow. Perhaps a new system could add clarity there.

Perhaps it could also continue to fly on the crushing course with visual perception and split the top grades. If that makes economic sense... it does sometimes and that I struggle to overlook

. Technical arguments can only add unbiased help. And that feel good.

There are a couple of comments about the other end of the scale on this thread ( esp. the second page: LINK). It would be great, I think, if someone can see a clear connection. For example, a unifying optimized and technically correct scale that could refine grades enough to distinguish the 'exceptional' colorless could lend authority to the other end of color that the current D-Z scale does not touch upon. Now that would be news, wouldn't it? Given the logarithmic mismatch between the D_Z scale and a corrected one, making an argument for split D hairs may be awkward, but a 'better than GIA D' and a precise split of GIA 'vivid' could easily go hand in hand. There may be only exotic use for the new top white area (although there is some use even now, with the IIa type having commercial appeal as a label), but splitting the yellow grades would probably surprise a smaller percent of the on-lookers. The current system seem to lack there - it seems unusual that color and colorlessness are graded not only on different scales but by different methods. I didn't jump up when the respective G&G issue came in - it seemed natural that GIA felt the need to write about it and there might be some economic argument in the respective 'gray' area to make as well.

Just my amateurish 0.2 - you know that.

 

[diamondsbylauren]

Hi Everyone!
I look at it this way:
It's easier to quantify the absence of color, as opposed to the prescence of color.
The scale is a great representation.
What's missing is presice calibration - in other words- D is one, and F is 100- If I was making such a chart, D would range from one to 2. E would range from 2.1 to 3.4 F would start at 3.5 till 5. G starts at 5.1 and goes to 7. H is 7.1-10. I is 10.1-16. J would be 16.1-24. 25 starts K- now, let's jump to the other end of the scale- Fancy Light Yellow would be say, 200. Fancy Yellow would be 300.
100 times the distance between D and E.


However,I agree with......yes, Garry.
Life, and diamonds are not linear.
I feel GIA's system to be a very good compromise of "statistically" grading the color of diamonds, while also taking into account that judging the color of something is always going to be subjective to some degree.

 

[Serg]

800 steps for yellow diamonds. Why are not 8.000?


What is color for color stone if all virtual facets have different color( and color for each facet depends from position diamond, ...)


Dave,

Please start from definition: What is color of polished diamond? ( I know only what is color for flat plane. I know what is GIA color for diamond by color gauge. But what is color for polished diamond without gauges and human?)

Then we can discuss scale.

 

[oldminer]

We are able to grade the color of diamonds now with technology combined with a database of what GIA has graded similar stones. It is based on 2 components; a numerical result with a spectrometer AND also on statistical modelling of sufficient previous GIA results.

The question is, is there any reason to re-make a working system?

It seems the fancy end of the spectrum could use the most help. The colorless range is reasonably a happy situation as is. Is this the consensus?

 

[valeria101]

Date: 10/14/2005 4:32:57 PM
Author: oldminer


It seems the fancy end of the spectrum could use the most help. The colorless range is reasonably a happy situation as is. Is this the consensus?

I can''t answer that one, but would take up the chance to ask: does this imply a consistent rule throughout the spectrum ?

It doesn''t sound impossible,

only somewhat different than the current practice (I am trying to imagine an index of visible color made with some saturation dependent weights so as to leave the colorless and near colorless not affected by hue etc and leave the face-up view relevant for the colored end as is etc.).

 

[Garry H (Cut Nut)]

Dave in ref to Sergey''s question - how would you grade the stone in figure 3 A Gems & Gemology Summer 2005.

Do you grade face up or as gIA do for D-Z stones - do you grade body color?

 

[diamondsbylauren]

Hey Serge- What a great point!
If we could definatively quantify the color if any particular facet of a diamond, it might not be exacly the same as every other facet.
In fact, how a diamond is veiwed makes a HUGE difference.
And you can not draw the conclusion that this is either consistent, or linear.

As Garry points out, GIA grades "colorless" diamonds from the back , and fancy color diamonds face up.
I will say that based on the diamonds we''ve seen in the U-V, W-X, and Y-Z ranges, GIA seems to grade these for "Face Up" color. Well cut Radiant stones in these grades do seem farly consistent as to face up- that is to say, Y-Z generally faces up darker than W-X- regardless of how the diamond looks from behind.

One of the reasons you rarely see Fancy Colored Round Brilliant Cuts is because they seem to make color wash out.
So if you look thru the pavillion ( from the back) a round can look mighty yellow- when you turn it over it looses that color.

The better radiant cut diamonds seem to do the opposite.

Dave- I guess I''m with the if it ain''t (horribly) broke, don''t fix it school.
When we discuss colored diamonds, the conversation is brings us to this fact: GIA is clinical in nature- that benfits the trade, because, ostensibly, clinical is impartial.

Is Fancy Orangey Yellow worth more than Fancy Yellowish Orange? GIA''s impartiality is benficial to everyone. If GIA was to assign values- like the lesser labs- it would destroy that impartiality.
By the same token, a GIA report can not tell you if it''s a good diamond. Even if it''s a D Flawless.

 

[Serg]

Date: 10/14/2005 4:32:57 PM
Author: oldminer
We are able to grade the color of diamonds now with technology combined with a database of what GIA has graded similar stones. It is based on 2 components; a numerical result with a spectrometer AND also on statistical modelling of sufficient previous GIA results.

The question is, is there any reason to re-make a working system?

It seems the fancy end of the spectrum could use the most help. The colorless range is reasonably a happy situation as is. Is this the consensus?

re: a numerical result with a spectrometer


What type a numerical result with a spectrometer do you want use?

Do you can receive correct absorption spectrum for round diamond cut?( absorption spectrum of diamond material)
I think, you can not do it.

Color each facet depends from absorption spectrum material , CUT( 3D model including size ), positions( diamond, lights and viewer) and other parameters.

 

[Regular Guy]

Date: 10/13/2005 10:09:22 AM
Author: oldminer


It probably would be most useful to have an even number of steps in each color. I suppose this would be some version of logarithmic. The problem is the unscientific spacing of the existing GIA color grade scale. It is not perectly spaced from color to color, especially in the H I J regions.

All your thoughts are appreciated....

Sorry to interrupt the high quality of this discussion, but was tempted by Dave''s comment to chip in...

I guess I had the impression that the current color system was Charlottesville-esque. That is...when you go to the University of Virginia in Charlottesville, you could look down an expansive field and see a row of columns. If you stand at the head of the row -- visually -- they look evenly spaced, and were in fact designed to look that way, from the point of view of the person standing at the head of the columns. Then, if you walk down the field, you see the columns are not evenly spaced at all, but are unevenly presented. Nevertheless, in the same way that an artist tries to render a picture in a realistic way, using artistic tricks, in the same way that Jefferson designed the columns to look evenly spaced from their head -- by in fact making them unevenly spaced down the line -- so too, or so I thought, the colors might have been separated so that visual demarcations effectively broke reasonably, from one color to another, even though digital renderings might score them differently.

If this was like saying there''s a nose on my face, apologies.

Regards,

 

[oldminer]

The color grading we are doing by machine is using light images captured from the side view. At this time, we are grading primarily D to M range diamonds, not into the lower end of the D-Z or fancy scale. We use many scans from a revolving stone as a signle view would not be sufficient.

Once we start to grade fancy color diamonds, we will use top views combined with side views, in all probability. In the attempt to give equivalent results to GIA grading our logic has been to use similar viewing positions.

I don''t have the G&G 2005 edition with me. What ind of a diamond was veing referred to?

 

[Garry H (Cut Nut)]

Here is the photo I referred to Dave.

I think it makes the point Sergey is pointing to.

Even if the stone is seen from the side there will be several different zones of varying color saturation.


I suspect an 'average result' would change if you recut the same material into a different faceting shape?

( I did 2 little cut and pastes from the area near the top left of the left side image to show the right side stone really is not digitally enhanced other than how the said they did)