What is possible in grading with computers?

[mrmedoes]

This is one of those hypothetical questions that has been really making me wonder for a little while now. Are there computers out there that can essentially "score" grades for diamonds? Now before anyone flies off the handle here about human touch and eye, I understand such things, but with such variances in grading from company to company (and person to person for that matter), I would think that there must be some sort of way for standardization. Theoretically, things such as color should be able to be graded to a very exact science. The same would go for inclusions, unless I''m missing something there – obviously “eye clean” would still be a relative term.

I do realize that the Sarin exists, but I''m thinking even beyond that. Is there already something major out there that I''m completely missing, or is technology just not ready for that yet? Obviously, I would figure there would be some major grumblings from anything of the sort due to cost, possible lost jobs, etc, but it would add some much needed legitimacy to grading.

What am I missing here or what is out there and in development? Things like this tend to interest me, so I figured I’d bring it up.

 

[RockDoc]

Date: 2/18/2005 9:48:57 PM
Author:mrmedoes

This is one of those hypothetical questions that has been really making me wonder for a little while now. Are there computers out there that can essentially 'score' grades for diamonds? Now before anyone flies off the handle here about human touch and eye, I understand such things, but with such variances in grading from company to company (and person to person for that matter), I would think that there must be some sort of way for standardization. Theoretically, things such as color should be able to be graded to a very exact science. The same would go for inclusions, unless I'm missing something there – obviously “eye clean” would still be a relative term.

I do realize that the Sarin exists, but I'm thinking even beyond that. Is there already something major out there that I'm completely missing, or is technology just not ready for that yet? Obviously, I would figure there would be some major grumblings from anything of the sort due to cost, possible lost jobs, etc, but it would add some much needed legitimacy to grading.

What am I missing here or what is out there and in development? Things like this tend to interest me, so I figured I’d bring it up.

Mr. Medoes...

Your question is very good.

I'll attempt to address it.

There are a lot of different categories that computers can do, but they are limited.

Let's start with ray tracing. Serg from Russia has developed Diamond Calc. There is in the works, or at least partially completed a 3D system where the diamond's measurements are imported in from a Sarin (provided the Sarin owner has the DIA PRO software, and Sarin's upgraded 3D version).

Holloways HCA also works on the principal of caluclated formulas. The ray tracing software which does depend on mathematical calculation has been done by reverse ray tracing. It should be forward ray tracing. Also the metrics to do this really accurately requires programming that is enormous, and calculating each stone might take a significant amount of processing time. As computers become faster and more powerful we will probably see developments in this area.

Clarity grading. The GIA and the AGS are probably going to make changes in their standards for grading. For sure proportion standards. How intricate these will be we'll know when they release the software and implement the changes to their labs - AGS LAB and Gem Trade Lab.

Equipment manufacturer's have been trying to automate clarity grading. But in order for it to become a standard in the grading community, GIA or AGS or both have to adopt such equipment and publish standards so the manufacturers can produce instrumentation that parallels their systems. There was one unit that would clarity grade produced, but the company never got it to market where the labs, jewelers and appraisers started using it.

Light performance grading. There is the new IMAGEM machine which Dave Atlas has. However, we don't know what its capabilities are yet. Dave has written a little about it. He has a machine installed but what it is actually capabile of and how it works hasn't been published as of yet.

Gem Ex - The Brilliance Scope - At this moment in time, it is the major player in town. They have had the machine out for years, and are constantly improving upon its capabilities. 700,000 diamonds have been imaged and are constantly being studied to improve the technology and gradings. See http://www.gemex.com

The ISEE2 machine developed in Israel is marketed by a diamond dealer who requires users to purchase at least some of their diamonds from that company. While I don't have a personal experience with it, I have heard the scintillation method is rather unusual but the brilliance system is sort of "all wet". Jonathan at good old gold has one, so maybe he will comment on his experiences with it.


The SAS 2000 made by Martin Haske. A very interesting machine which is basically works with a photospectrophotometer. It has a new laser attachment. Martin has some very advanced software with this machine that is linked to the individual spectrum analysis of the chemical content of stones. It is rather complex and can detect treatments such as hthp and irradiation of colored diamonds. It also analyzes colored stones, and can color grade diamonds too. There is a lot to know about this so I am linking Martin's site here http://gis.net/~adamas/

There are some other units which can color grade, but they vary a bit, and are not toally accurate in some cases. The leading machine is a Gran machine which was developed in Israel. The company was sold to Sarin. It is a good back up, as if there is a difference between the color grade of a grader with the proper lighting and DIAMOND Master Comparison stones, it alerts the grader to possibly review his grading. So it is a good back up, but not the primary instrument for this.

Color grading is a lot more complex than you might imagine.

There is a great tape available from the HRD Lab in canada that pictures and describes advancements in computer driven machines that analyze rough diamonds. It is photographed well and describes and pictures laser instrumentation for sawing diamond crystals, and is a great visual depiction of the happenings in Antwerp.It also has photos of the "octopus" and automatic cutting machine too. I think the cost is around $10.00 and a excellent visual source of learning about developments in Antwerp.

I am sure there is more indepth news about things to come. One of them is a machine that is supposed to be available shortly to help with the hpht detection. It is supposed to become available this coming Spring.

So there is a lot of new things to come, and probably a couple of years from now, even a lot better and more efficient equipment to be introduced too.

I agree with you that a standard that would be equal between everyone would be an improvement, but I am not sure that will ever happen. But we'll see as time goes by. A lot of effort is being made to improve many areas of the varied specialties of the industry.

Hope this helps

Rockdoc

 

[strmrdr]

not much.
The grading standards are way too subjective for a machine to gain wide acceptance.
Lets take color:
Someone with a lot of money could offer 5 million dollars for a diamond that 5 top labs and 10 appraisers all rate a G color and be confident in never having to pay out anytime soon.
They cant even agree what is the best light to grade them under much less anything else.

 

[RockDoc]

Date: 2/19/2005 12

1

1 AM
Author: strmrdr
not much.
The grading standards are way too subjective for a machine to gain wide acceptance.
Lets take color:
Someone with a lot of money could offer 5 million dollars for a diamond that 5 top labs and 10 appraisers all rate a G color and be confident in never having to pay out anytime soon.
They cant even agree what is the best light to grade them under much less anything else.

I am trying to interpret your meaning here..... Maybe I''m dumb, but I can''t seem to grasp the concept of the above.

What''s the point of 5 million - 5 top labs, and 10 appraisers?

"and never having to pay out anytime soon"?

The best I can make of this is that your saying that grading diamonds is so subjective, that there isn''t a credible grading standard? Am I right?


Can you clarify that please.....

Rockdoc

 

[strmrdr]

Date: 2/20/2005 12:23:37 AM
Author: RockDoc

Date: 2/19/2005 12

1

1 AM

Author: strmrdr

not much.

The grading standards are way too subjective for a machine to gain wide acceptance.

Lets take color:

Someone with a lot of money could offer 5 million dollars for a diamond that 5 top labs and 10 appraisers all rate a G color and be confident in never having to pay out anytime soon.

They cant even agree what is the best light to grade them under much less anything else.

I am trying to interpret your meaning here..... Maybe I'm dumb, but I can't seem to grasp the concept of the above.


What's the point of 5 million - 5 top labs, and 10 appraisers?


'and never having to pay out anytime soon'?


The best I can make of this is that your saying that grading diamonds is so subjective, that there isn't a credible grading standard? Am I right?



Can you clarify that please.....


Rockdoc

Sorry i guess it was as clear as mud lol.

The main point is that there isnt one industry standard on just what color a G colored diamond is there are many.
Finding 5 labs and 10 appraisers that all agree one diamond is a G color is pretty much impossible.
Getting everyone on the same page and staandard will be next to impossible because the current system makes a lot of industry heavyweights more money.
The only way I see it happening is goverment regulation and they would mess it up worse than it is.
So even if someone develps a 100% accurate and repeatable computer controlled machine to measure color the problem becomes whose standard do you calibrate too?
Which standard will everyone accept?
Will the heavyweights give up their golden egg?

I hope that is a little more clear....

 

[Richard Sherwood]

Date: 2/20/2005 12:33:12 AM
Author: strmrdr

The main point is that there isnt one industry standard on just what color a G colored diamond is there are many.

Actually, that's incorrect Storm. The GIA originated the color grading system with a set of master stones clearly delineating the grades. These grades are an absolute, with a range alloted between the adjacent grades (a Gran colorimeter for example illustrates the range with 5 "clicks"). Subsequent sets of master stones are graded using the original set's delineation as their standard.

Finding 5 labs and 10 appraisers that all agree one diamond is a G color is pretty much impossible.

Again, incorrect. The fact is that the majority of credible labs and precision appraisers would be in agreement. It would be the minority that would fall high or low of the grade, and the majority of them would be within the industry accepted one grade tolerance.

Getting everyone on the same page and staandard will be next to impossible because the current system makes a lot of industry heavyweights more money.

Not true. People said the same thing when the GIA grading system came out. Before that you had all kinds of grading abuse with everybody using their own terminology. The GIA system established a standard, which gradually became accepted and more & more adhered to.
The precision of grading is increasing with better trained gemologists and advanced equipment, such as colorimeters and spectrophotometers. Additionally pressure generated by consumer awareness is creating tighter and more precise grading to satisfy the demand for repeatability.

So even if someone develps a 100% accurate and repeatable computer controlled machine to measure color the problem becomes whose standard do you calibrate too?

In this area the GIA has always been the leader which the rest of the world's labs have followed. My guess is that this will remain so for quite some time to come. There might be a team of labs which will be an industry vanguard, but their step will probably match the GIA's in cadence.

Which standard will everyone accept?

The GIA's, albeit cloned and filtered throughout the world diamond industry.

Will the heavyweights give up their golden egg?

I don't really understand what you're talking about here. How will advanced precision grading hurt the major lab's profits in any way? If anything, it will increase it.

No, the incentive is to progress, not to remain stagnant or fight progress. It might not be as fast as we want it, but it will happen. Do you have any idea of the amount of money spent by the GIA for research and development? Its big bucks.

 

[strmrdr]

Richard I hear what your saying but we both know it dont work that way in practice.
An G colored diamond as graded by egl might or might not get a G grade by the GIA.
Same with igi vs gia.
Sometimes GIA itself cant even agree with itself if the stone is sent in twice.
Here is my proof:

http://grading.pricescope.com/

The heavyweights Im talking about are not the labs themselves but some cutters and suppliers in the supply chain.

 

[Richard Sherwood]

Storm, if anything this comparison illustrates the consistency shown between labs in their grading:

In half the cases AGS clarity grading was one grade stricter than GIA. In two thirds of cases EGL USA color grading was one grade softer but clarity grading in a third of cases was stricter than GIA. Industry opinions were confirmed and by large each lab graded consistently to its own apparent standard, considering the high degree of human subjectivity involved.

In other words, grading was never more than the industry accepted tolerance of one grade different between these three labs.

On top of this is the natural correction implented by the pressure of the market, in which the discounts given between the different certs invariably placed the stones at their correct market level.

A check, and doublecheck. And it''s only getting tighter and tighter...

 

[oldminer]

The Imagem device does measure diamonds with extreme accuracy. We will shortly be issuing full reports on diamonds that are totally automated, though reviewed by a competent gemologist before printing. All physical measures will be to three decimal points with complete repeatability. Three seems totally adequate although the device is far more accurate.
This is much more accurate than Sarin or Ogi presently.

It measures light behavior: Light return (Brilliancy), Scintillation (Sparkle) and Pattern (Intensity). It does not calculate these facts, but physically measures each individual diamond with whatever oddity of cut it may have.

It measures color grade scientifically and with statistics accurately states the proper GIA color based on the centroids of what GIA has called the color of similar stones. There is no attempt here to remake our GIA color grading system, but what is being done is to completely remove the subjectivity from the GIA system. Every time the same diamond gets graded, the same grade will result. Admittedly, there are diamonds on the bordlines of one grade to another. There can be arguments over which grade is proper, but when a device gives objective results these borderline cases become a meaningless argument. I don't think we want to further divide the color grades. There are plenty already.

The Imagem device gives repeatable strength, and wavelength (color) of long wave UV fluorescence. Not subjective, but measured.

ImaGem has a clarity module which automatically finds and plots inclusions in a repeatable manner. This will look a little different than the loose GIA, AGS or EGL style of plotting, but a clarity plot should be consistent, meaningful and indicative of what is in the diamond. This is done well by ImaGem. Imagem also calls the clarity grade and will become more reliable as more stones are processed. Nothing is more complex than the myriad of rules and strategies for calling a
correct clarity grade. It can, and wil,l be done. Automated plotting is a big step. The total automation of most clarity grading will not be far away.

Lastely, ImaGem permanently recognizes any diamond processed and can register the identity of the stone either in a local database or in the ImaGem International database. This is without any Gemprint process or laser inscription marking. A diamond once processed can be forever recognized, safely traded, sent on memo, or left for jewelry repairs without fear of a changed stone being returned.

Obvously, this technology will leave a few strange diamonds unable to be machine graded. This is what an expert gemologist will have to address. The vast majority will grade very nicely with only limited supervision and expertise. For only a few million more dollars, additional diamonds could be graded automatically, but the financial benefit of even more advanced technology is outweighed by the extreme expense involved. The present work has been in the millions already and has taken nearly 10 years.

No doubt, for a while there will be skeptics. New technology goes through a stage where change creates doubts. We will be pleased to offer a few samples of the full report output to qualified dealers in the coming weeks and months. Right now, we are ready to provide light behavior/cut grade reports which are suitable to accompany all existing lab reports regardless of which lab was utilized. These are being made for consumers and the trade. Dealers who wish to have a few sample stones run to examine the results are more than welcomed.

Quality input from knowledgeable dealers and gemologists will only improve the ImaGem product. We look for constructive advice, as we go forward.