Do you think diamond grading will ever become 100% automated?

[Adylon]

We all know there are subjective diamond grading characteristics (clarity, color, polish, symmetry, etc...). Sure there are machines and tools that help but the final call always comes down to a human's interpretation to what he/she thinks it deserves.

So my question is do you think the major labs would ever adopt machines or tools specifically designed to measure these characteristics instead of just human review? Why don't they all use a colorimeter for example or some other similar device? That one has double the precision (it measures +/- 1/2 a color grade loose) plus of course you benefit with the repeatability of the machine and lack of human error, so wouldn't going to a 100% automated/calculated system be preferable? Certainly they can create a machine that could measure clarity if they really wanted to.

The only reason I can think of that they choose not to go this route is because it would empower the cutters, manufacturers and retailers to purchase the equipment and grade diamonds on their own and reduce their need to rely on the labs for definitive grading. So therefore human grading perpetuates their business model and maintains their authority status which is why they do not automate it.

 

[diagem]

Date: 2/28/2007 2:08:57 AM
Author:Adylon
We all know there are subjective diamond grading characteristics (clarity, color, polish, symmetry, etc...). Sure there are machines and tools that help but the final call always comes down to a human''s interpretation to what he/she thinks it deserves.

So my question is do you think the major labs would ever adopt machines or tools specifically designed to measure these characteristics instead of just human review? Why don''t they all use a colorimeter for example or some other similar device? That one has double the precision (it measures +/- 1/2 a color grade loose) plus of course you benefit with the repeatability of the machine and lack of human error, so wouldn''t going to a 100% automated/calculated system be preferable? Certainly they can create a machine that could measure clarity if they really wanted to.

The only reason I can think of that they choose not to go this route is because it would empower the cutters, manufacturers and retailers to purchase the equipment and grade diamonds on their own and reduce their need to rely on the labs for definitive grading. So therefore human grading perpetuates their business model and maintains their authority status which is why they do not automate it.

These Colorimeter or other like devices out there are NOT that accurate (at-all)!!! Although I am pretty sure the Laboratories have them in use...
If i am not mistaken, they issue color grades based on "substance memory"!!!

As far as clarity... for example how would a "device" categorize a specific feather???? (there are feathers of same characteristics that can be called/graded in the range of SI2 to VS2! based only on their location in the Diamond.) Do you think a device can be a proper judge on these type of issues???

 

[pricescope]

Yosef, as long as a machine is operated by human there is always a room for both - honest (tuning, not cleaning properly, poor maintenance, plain damness...you name it) and dishonest mistakes in measurements.

 

[denverappraiser]

I’ve never seen an industry where the powers that be were interested in protecting the jobs of the workforce and I see no reason to believe that the labs will be different. A tool that could allow graders to accurately grade diamonds significantly faster and/or consistently would be enormously popular with the owners of the labs. If that puts some graders out of work, so be it.

Tools continue to improve and the labs are pretty good about remaining on the cutting edge of using the best available equipment. Imagem professes to do exactly what you describe and is actively selling their equipment to manufacturers in India.

Theoretically, each inclusion could be assigned a numeric score based on it’s relative size, makeup, location, relief etc. and a formula could be created that would distill these scores down to a total clarity grade. This would require a pretty clever scanner but I see no reason that a tool can’t take into account the inclusions location relative to both the stone and the other inclusions. Machines are far better at this sort of thing than people.

I have my doubts that this will spell the end of the labs but stranger things have happened. 50 years ago, the way almost everyone bought a diamond was to rely on the integrity of the jeweler to supply a stone of a particular grade. Labs didn’t exist. This has evolved to where consumers now prefer to put their trust in a remote lab that they see as an expert. The business of labs has become big business indeed and supplanting them as the trusted source for diamond information would take more than just a tool. Tools require calibration. Tools can be tampered with. Tools can be counterfeited. This means that trust really is with the tool operator, not with the tool itself. Are consumers really ready to put their trust in the manufactures to accurately report what they produce? This seems unlikely. Will it move back to the retailer, who could then buy the machine? Maybe, but I doubt this as well. At least for the foreseeable future, I think customers will continue to like the idea that their stone has been graded by an ‘independent’ service and that their reporting will continue to be part of the value of the stones. What equipment the labs use will be of academic interest to the diamond junkies and people in the trade but most people neither know nor care.
Neil Beaty
GG(GIA) ICGA(AGS) NAJA
Professional Appraisals in Denver

 

[tanalasta]

I remember on Rhino''s goodoldgold site that the newest generation of colour grading machines are actually quite accurate.

http://www.goodoldgold.com/4Cs/Color/DiamondColorimeter/

"Q1. How accurate is the colorimeter?
A1. Very accurate. A properly calibrated Colorimeter grades extremely accurate. Just about all of the diamonds we have graded and then double checked against a set of master stones plus confirmed with lab reports demonstrate to us the accuracy of the Colorimeter. This is the 3rd generation Gran Colorimeter and we have been using it since its inception. The latest model, the DC3000 has proven to be the best and most accurate machine produced by them to date. We have had some problems with past models but so far no probs with their latest generation."

One day, a laboratory with a helium scanner, the appropriate software to create 3D recreations of the diamond and perhaps also detect and plot inclusions and measure light performance ... could exist one day!

Not sure how a machine would grade ''polish''.

 

[RockDoc]

Equipment is very nice to have BUT...............

You have all the issues that Pricescope wrote.... calibration, deception, carelessness in operation, level of expertise of the person operating the machines.

The proportion machines can do "tasks" in a few moments, that the human eye doing it manually either couldn''t or would take hours to accomplish. This is still a technology that will improve in the future relative to tightening its accuracy and tolerances.

As for clarity, machines already exist that will assign clarity grades, however they do not take the place of human checking, which has to be done. There have been lots of attempts to "machine" grade clarity, but none have really "stuck".

Machines are a good "back up", but trained human supervision is very necessary. Machines, also unemotionally make their conclusions, which do ( at least at the moment) do not understand or are capable of measuring and assessing BEAUTY, Attractiveness, and the like. They are good "indicators" but they still need review by expert eyes. Will equipment developers come up with machines that have human logic? Maybe, but at least at this moment in time, I would doubt it.

The expert eyes still have the final trump card.

Maybe it will change in the future, but at least for now that''s how I see it.

Rockdoc

 

[Nicrez]

Some labs are currently automating more and more. Consumers can not know the extent of these changes, because many of them are trade secrets, and one lab can not let out it's secrets, should another lab find out. The conusmer would be the absolute last to know, so as for labs going automated, it's certainly going that way, and you can bet your paycheck that once the machines are more accurate and consistantly, they will replace each human grader. Look at the Sarin. Very few graders use gauges to measure anymore, and the colorimeter IS used in some labs, but again tha accuracy and consistancy is not there yet. Somehow the human eye (even when varying from person to person) still consistantly gets the same answer because the variables for a machine response have to be perfectly controlled, and that is not necessary for human as much.

Diagem and Denverappraiser, there are machines that are being worked on that use a high frequency light (possibly laser) that goes through the stone and detects for distortions in the crystal lattic. Such distortions would be maped based on shape and size, but the human element would still have to ratify and agree with these "inclusions". The mapping technology is not too far away. Not too far away at all...

 

[Paul-Antwerp]

Date: 2/28/2007 10:00:34 AM
Author: tanalasta
I remember on Rhino''s goodoldgold site that the newest generation of colour grading machines are actually quite accurate.

http://www.goodoldgold.com/4Cs/Color/DiamondColorimeter/

''Q1. How accurate is the colorimeter?
A1. Very accurate. A properly calibrated Colorimeter grades extremely accurate. Just about all of the diamonds we have graded and then double checked against a set of master stones plus confirmed with lab reports demonstrate to us the accuracy of the Colorimeter. This is the 3rd generation Gran Colorimeter and we have been using it since its inception. The latest model, the DC3000 has proven to be the best and most accurate machine produced by them to date. We have had some problems with past models but so far no probs with their latest generation.''

Let me tell you about my experience with the Gran DC3000. I too found it rather accurate, when using it on our rounds and our princess-cuts.

However, a few weeks ago, we were using it on princess-cuts of rather low cut-quality. And I just had to stop the experiment, because the machine was giving me results which were 3 or more colours off.

My take on this: the Gran is a good colour-machine, if used correctly and with attention, but probably only on stones with a relatively high light-return. Considering that most of GOG''s stock is of relatively high cut-quality, I can imagine that they consider the machine to be perfect. But my experience shows that it is not ready for lab-use yet.

Live long,

 

[Adylon]

Thanks for everyone's responses... I guess I hadn't thought about operator error, calibration issues, etc. For +/- 1/2 a grade I can see where this could still be a concern, but let's imagine you could measure color as well as carat weight. If a jeweler sells you a diamond as 1.01cts .. you can assume maybe that it's only 1.00Xcts but if it was checked to be 0.940cts this is OBVIOUSLY fraud because no legit diamond scale can be that innacurate. Maybe in the future we could see colorimeters as accurate as say F.01, F.02, ... F.98, F.99, G.00, etc. then perhaps people would have faith in them?

I don't see why characteristics such as feathers or their relative placement, etc. can't all be programmed with logic into a clarity machine eventually. There are rules regarding the grading of clarity that is taught to gemologist students, so couldn't these same rules be programmed in a machine? It's all in the programming, the better the programming, the more logical rules the more accurately the machine can extrapolate and better interpret the rules for each diamond to make judgement calls.

I guess what really bothers me most about grading is how the scale is so finite and absolute when it really isn't. I think people would enjoy shopping for diamonds more if they hade better faith in what they were buying was accurately graded.

 

[diagem]

Date: 2/28/2007 12:48:56 PM
Author: Adylon
Thanks for everyone''s responses... I guess I hadn''t thought about operator error, calibration issues, etc. For +/- 1/2 a grade I can see where this could still be a concern, but let''s imagine you could measure color as well as carat weight. If a jeweler sells you a diamond as 1.01cts .. you can assume maybe that it''s only 1.00Xcts but if it was checked to be 0.940cts this is OBVIOUSLY fraud because no legit diamond scale can be that innacurate. Maybe in the future we could see colorimeters as accurate as say F.01, F.02, ... F.98, F.99, G.00, etc. then perhaps people would have faith in them?

I don''t see why characteristics such as feathers or their relative placement, etc. can''t all be programmed with logic into a clarity machine eventually. There are rules regarding the grading of clarity that is taught to gemologist students, so couldn''t these same rules be programmed in a machine? It''s all in the programming, the better the programming, the more logical rules the more accurately the machine can extrapolate and better interpret the rules for each diamond to make judgement calls.

I guess what really bothers me most about grading is how the scale is so finite and absolute when it really isn''t. I think people would enjoy shopping for diamonds more if they hade better faith in what they were buying was accurately graded.

What are the rules?

 

[Adylon]

When I mean rules I guess I mean a database of some real examples to compare A to B one by one just like a human would learn. If a computer could reference say 50,000 images or 3D maps of included diamonds, their inclusions relative positions and size in a database and what they were graded by a human, and compare them all to what it's scanning, why couldn't it be programmed to calculate clarity similar to the way a human would calculate it? My background wasn't jewelry at first, it was mechanical and electrical engineering. I haven't taken any formal gemology courses either, I'm sure you guys know better then I. But the engineer in me just wishes there was a better tool to standardize these types of measurements.

 

[oldminer]

There is little doubt that in the future a large percentage of diamonds will be graded by machine and only checked over by a trained person who rarely needs to change any results. There are odd stones which may defy the initial automation technology, but more and more of these issues will become solved problems over the years.

Whoever thought a Lexus could parallel park itself? Things we can''t imagine are going to happen.