Consumer advisory: GIA Cut Grade Rounding Problems
- Thread starter JohnQuixote
- Start date
- Status
- Joined
- Aug 15, 2000
- Messages
- 18,479
Marty they did +70,000 observations and the dome only constituted about 20,000 of those observations
I did count the dome observertions here once before http://www.gia.edu/pdfs/cut_table_1.pdf
You can also see that the two largest +5,000 sets were essentially focusing on the 46 degree obstruction (one was 23-0 black and the other one was 22.5 degrees to the horizon black
- Joined
- Aug 15, 2000
- Messages
- 18,479
Marty all the way through you can see that they were struggilng to find some metric to explain what the observers were seeing. They clearly would have liked to have some calculation system to predict the boundaries accurately.
I do not believe they ever found that method. If they did then they certainly never mentioned it in that Foundation article.
I can understand Marty that as a scientist you are looking for reproducable techniques to try to test their results.
I hope you can understand though that what I am saying is most likely to be correct.
GIA conducted about 50,000 observations, mostly in light boxes, and mostly with trade participants. From this data they built a parametric grading system.
Because of the lack of a "metric" they are unable to grade stones with indexed (painted and dug) faceting because that would entail conducting another survey with additional various poprotions that would require about twice as many samples and analysis again as the first round.
In conclusion, GIA have a parametric grading system that works the same way that HCA does, except it penalizes minor facets if they are outside a very wide range of unusual minor facet variants. They have also included Sym, polish and girdle thicknes etc (which I too have on the desk top HCA, but it is too much to expect consumers to work out girdle thickness from many different lab systems)
I do not believe they ever found that method. If they did then they certainly never mentioned it in that Foundation article.
I can understand Marty that as a scientist you are looking for reproducable techniques to try to test their results.
I hope you can understand though that what I am saying is most likely to be correct.
GIA conducted about 50,000 observations, mostly in light boxes, and mostly with trade participants. From this data they built a parametric grading system.
Because of the lack of a "metric" they are unable to grade stones with indexed (painted and dug) faceting because that would entail conducting another survey with additional various poprotions that would require about twice as many samples and analysis again as the first round.
In conclusion, GIA have a parametric grading system that works the same way that HCA does, except it penalizes minor facets if they are outside a very wide range of unusual minor facet variants. They have also included Sym, polish and girdle thicknes etc (which I too have on the desk top HCA, but it is too much to expect consumers to work out girdle thickness from many different lab systems)
- Joined
- Aug 15, 2000
- Messages
- 18,479
- Joined
- Aug 15, 2000
- Messages
- 18,479
Also note on the chart above - how few poor stones they studied.
This is the achillies heel of all the systems that i know of - they only run the checks on the good to best stones - and you can not verify and validate a system this way.
I know of a test that was run on Brilliancescope with one of the actual stones from the GIA study - a shocker that scored very high on BS. The person who ran it has gone very very quiet about technologies.
Dave if you are reading this - please do it on Imagem - test all the worst stones you ever saw - and all the unusual stones - ones with great fire and dead brilliance etc
GIA have still never accounted for the stones that they got off the scale brightness for with shallow crowns.
Some of those stones have great fire too. Any good system should be able to predict or measure this.
This is the achillies heel of all the systems that i know of - they only run the checks on the good to best stones - and you can not verify and validate a system this way.
I know of a test that was run on Brilliancescope with one of the actual stones from the GIA study - a shocker that scored very high on BS. The person who ran it has gone very very quiet about technologies.
Dave if you are reading this - please do it on Imagem - test all the worst stones you ever saw - and all the unusual stones - ones with great fire and dead brilliance etc
GIA have still never accounted for the stones that they got off the scale brightness for with shallow crowns.
Some of those stones have great fire too. Any good system should be able to predict or measure this.
- Joined
- Aug 15, 2000
- Messages
- 18,479
i will check in
off on a trip
off on a trip
adamasgem
Brilliant_Rock
- Joined
- May 23, 2003
- Messages
- 1,338
Belle
What is NOT CLEAR in reading the GIA article, is whether or not the Brightness and Fire Metrics placing the relative positions are based on:
1) a symmetrical stone assumption or
2) whether they are based on the actual asymmetrical stone.
It would be interesting to see the Sarin scans and see just how good or bad the stone really are, symmetry wise
It behooves GIA to add the uncertainty of their metric positioning if the first case is true, which I suspect it is, since they have never published anything relating to assymmetry other than say they did it, but only data based on "rounded" averages and/or symmetrical stone models.
One does note the change in the GIA parametric "definitions" of their same test stones across the three G&G articles. Noteable is the rounding occuring from the Fire article to the last 2004 G&G article..
AGS uses the "actual" assymetrical stone scan (however errored) and doesn''t force symmetry into the process.
belle
Super_Ideal_Rock
- Joined
- Nov 19, 2004
- Messages
- 10,285
of course it's not clear!
gia may (double italicize and underline 'may') have started this process with the intention of getting good scientific results but there is no way around it...they %*#* up in that regard.
gia is the only entity with the funding capable of taking on such a project but unfortunately the scientific and accurate results would only serve to harm the many that are contributing to the funding in the first place.
there was so much potential here too.
Belle wrote :
gia is the only entity with the funding capable of taking on such a project but unfortunately the scientific and accurate results would only serve to harm the many that are contributing to the funding in the first place.
there was so much potential here too.
(*•.¸(`*•.¸¸.•*´)¸.•*)
¨`•.¸ *belle*¸.• ´¨
(¸.•*''(¸.•*´ `*•.¸)*•.¸)
Belle while you could have some deeply "smarts" here in reading between the lines... it could be worse!!!!
Someone could have gotten Jack Abramoff to do the "lobbying" at GIA on behalf of those who would want such a system in place. Or maybe funded someone with billios of government grant funding. Certainly we''ve funded far "sillier" subjects through federally subsidized grants.
Rockdoc
adamasgem
Brilliant_Rock
- Joined
- May 23, 2003
- Messages
- 1,338
The grapevine humormill emmanating from Tuscon was, that they had already hired the same PR firm to handle Certifigate as Bill Clinton did with Monicagate.
They already had the government subsidised funding in place given their tax exempt status. That means you, I and everyone else paid for it, and consumers are going to "pay for it" in the future.
- Joined
- Sep 3, 2000
- Messages
- 6,695
Belle wrote :
gia is the only entity with the funding capable of taking on such a project but unfortunately the scientific and accurate results would only serve to harm the many that are contributing to the funding in the first place. there was so much potential here too.
Actually, the GIA and AG'S are not the only entities with sufficient funding to have come up with the best, or the most correct, grading principals for light behavior grading standard creation. The firm I have been associated with for a number of years has a better, more scientific and unbiased process for grading diamond light performance. Several million dollars has been spent and it was NOT taxpayer's money, either. We are beginning this coming week to grade for light behavior over 26,000 diamonds; round, princess and marquise for a sight-holder in India and an American retail chain client. Every one of these stones will soon appear in the US market and we have many thousands more coming in over the next several months. Several other sight-holders are looking into this technology and process. You'll be surprised in coming weeks who else is getting involved with using this technology. Innovators and early adopters never rest. We'll see how it all plays out, won't we?
As more vendors take a long look at the available systems for grading diamond cut, I believe many will find excellent work has been done in this field, but not where one might have expected it to come from. This is typical of how a small technology company can make a breakthrough into a large, well established field which is lumbering along in the false belief that they have all the answers and they know all the players.
- Joined
- May 3, 2001
- Messages
- 7,516
Gary,
What are you saying here. What was the result and why did it make some one very very quiet?
This is a curious statement and it means nothing to me as you did not share the results versus the expected results or any of the information that would make it meaningful. Please share more.
Wink
- Joined
- Sep 3, 2000
- Messages
- 6,695
Marty, the point of discussion is to argue over points which are meaningful. The BIAS issue of importance evolves around what the biggest client's of GIA might have wanted, or what the users of ISEE2 might wish to promote in a specific brand. The bias in a lighting model is something to question, but it is a different sort of bias since all diamonds measured with this equipment get the same lighting model. Of course, we feel that the lighting model gives a diamond a chance to act in a normal manner. All these years were not spent in the attempt to make measures in a weird or unusual lighting environment where our own eyes would see no correlation to the grade given. In fact, the opposite was done. Very well respected observers with expertise in diamond cut were polled on how what they saw correlated to the grading provided. These devices need to produce results that meet our visual expectations. In due course, people on the outside of this business enterprise will be free to make these judgments, too. If people disagree, then the attempt won't be successful, but if the eyes of dealers, retailers and consumers agree with the results, then we have a good and fair metric which is can be used to grade light performance.
The correlation of cut grade to actual performance is not a big problem as the terms of performance grading are rather logical.
Brilliance: a measure of a stone's overall strength of light return that represents its average light return in the face-up position. This measure is arrived at by computing the mean gray-scale value of all pixels within the face-up view of a diamond.
Sparkle: a measure of those spangle-like flashes of reflected and refracted light that gives diamonds what might be called "kick" or “life”. The greater the number of these flashes, the higher a stone's sparkle. This measure is arrived at by measuring the standard deviation in the gray-scale value of the light return within the face-up view of the diamond.
Intensity: a measure of the number and strength of contrasting light-dark areas in the face-up view of a diamond that give it vitality and character. One may call this “contrast”. Generally, the greater the stone's symmetry the higher its intensity will be. However, other types of less symmetric cut may score well on this scale as well. This measure is arrived at by calculating a ratio of bright pixels to the total number of pixels within the face-up view.
These three measures, when combined with very simple physical measures which indicate qualities of durability, finish, and spread (size), known as DFS, will give everyone involved the knowledge of performance and cut craftsmanship quality that the industry and consumers are wanting to know. With these metrics we can grade any shape of faceted diamond. It is not a big deal to introduce grading of standard fancy shapes or for differently faceted, Branded cuts. It is a fair system, in that it does not favor any cutter, seller or maker. It measures light based on simple metrics, not convoluted predictions from rounded off, questionable data with too many variables tossed in just to muddy the waters. The more variables measured, the less reliable the end result. The more machine error and rounding in the measure of variables, the greater yet the error or lack of meaning to the end result. We are not suffering from any of this predictive problem. We agree that predictive devices will aid cutters, but direct measures reduce the variables to a very few, allow rapid results and pretty much eliminate errors from rounding or machine error.
I also know that few people have seen this at work. There is bound to be doubt and questions. It is part of what must be expected and I have faith we will be able to show the light behavior system at upcoming trade events beginning with the March 5 to 7th JCK NY Invitational Show in NYC.
The correlation of cut grade to actual performance is not a big problem as the terms of performance grading are rather logical.
Brilliance: a measure of a stone's overall strength of light return that represents its average light return in the face-up position. This measure is arrived at by computing the mean gray-scale value of all pixels within the face-up view of a diamond.
Sparkle: a measure of those spangle-like flashes of reflected and refracted light that gives diamonds what might be called "kick" or “life”. The greater the number of these flashes, the higher a stone's sparkle. This measure is arrived at by measuring the standard deviation in the gray-scale value of the light return within the face-up view of the diamond.
Intensity: a measure of the number and strength of contrasting light-dark areas in the face-up view of a diamond that give it vitality and character. One may call this “contrast”. Generally, the greater the stone's symmetry the higher its intensity will be. However, other types of less symmetric cut may score well on this scale as well. This measure is arrived at by calculating a ratio of bright pixels to the total number of pixels within the face-up view.
These three measures, when combined with very simple physical measures which indicate qualities of durability, finish, and spread (size), known as DFS, will give everyone involved the knowledge of performance and cut craftsmanship quality that the industry and consumers are wanting to know. With these metrics we can grade any shape of faceted diamond. It is not a big deal to introduce grading of standard fancy shapes or for differently faceted, Branded cuts. It is a fair system, in that it does not favor any cutter, seller or maker. It measures light based on simple metrics, not convoluted predictions from rounded off, questionable data with too many variables tossed in just to muddy the waters. The more variables measured, the less reliable the end result. The more machine error and rounding in the measure of variables, the greater yet the error or lack of meaning to the end result. We are not suffering from any of this predictive problem. We agree that predictive devices will aid cutters, but direct measures reduce the variables to a very few, allow rapid results and pretty much eliminate errors from rounding or machine error.
I also know that few people have seen this at work. There is bound to be doubt and questions. It is part of what must be expected and I have faith we will be able to show the light behavior system at upcoming trade events beginning with the March 5 to 7th JCK NY Invitational Show in NYC.
- Joined
- Aug 15, 2000
- Messages
- 18,479
Good point Marty.
GIA have a Helium scanner.
Dear GIA Cut Study team, and G&G,
Please scan the stones in your collection and place the 3D scans in you data depository so that you can fully participate in peer review.
- Joined
- Aug 15, 2000
- Messages
- 18,479
Sorry Wink, I really wish I could - but it would betray confidences.
You know I would if I could.
Email all the people you ever met at Vegas who participate on this board from the trade and see if you can get them to share it.
RE: The correlation of cut grade to actual performance is not a big problem as the terms of performance grading are rather logical.
~~~~~~~~~~~~~~~~~~~~~~~~
Dave,
They may be logical to some people, but not to all. Sometimes I feel like the little pig in the brick house. Cut grading (rubric style of catagorizing) is not nearly precise enough to determine obvious cut differences from diamond to diamond. Add to this the fact that angle averages are used even adds to the sloppiness. Then add to this that the most important element of facet straightness is ignored puts such grading systems on shakier ground.
Performance grading in IMHO is itself on shakey ground. You already have performance sellers who quit selling a brand because of the non-correlation of what their customers see with their naked eye and the performance viewing machines.
My 28 years experience in the diamond industry as a cutter to retail seller and everything in between has taught me that the lay person is very capable of determining on their own what diamond looks good, but needs to be told which diamond is cut good.
It''s in stone: All diamonds that look good are not cut good. All diamonds that are cut good do not look good. Cut and Performance need to be separated.....just like a siamese twins.
Don''t underestimate the consumer....regardless of what level they are on. Anyone with eyesight can look at a diamond and form an opinion. It doesn''t matter to them if their opinion is the same as that of a viewing device, and this may well kill a sale. It''s the the work of the cutter that is the mystery to them, (and I have to say to many in the trade) and it''s the work of the cutter that needs to be related in simple terms between buyers and sellers.
Bill Bray
Diamond Cutter
~~~~~~~~~~~~~~~~~~~~~~~~
Dave,
They may be logical to some people, but not to all. Sometimes I feel like the little pig in the brick house. Cut grading (rubric style of catagorizing) is not nearly precise enough to determine obvious cut differences from diamond to diamond. Add to this the fact that angle averages are used even adds to the sloppiness. Then add to this that the most important element of facet straightness is ignored puts such grading systems on shakier ground.
Performance grading in IMHO is itself on shakey ground. You already have performance sellers who quit selling a brand because of the non-correlation of what their customers see with their naked eye and the performance viewing machines.
My 28 years experience in the diamond industry as a cutter to retail seller and everything in between has taught me that the lay person is very capable of determining on their own what diamond looks good, but needs to be told which diamond is cut good.
It''s in stone: All diamonds that look good are not cut good. All diamonds that are cut good do not look good. Cut and Performance need to be separated.....just like a siamese twins.
Don''t underestimate the consumer....regardless of what level they are on. Anyone with eyesight can look at a diamond and form an opinion. It doesn''t matter to them if their opinion is the same as that of a viewing device, and this may well kill a sale. It''s the the work of the cutter that is the mystery to them, (and I have to say to many in the trade) and it''s the work of the cutter that needs to be related in simple terms between buyers and sellers.
Bill Bray
Diamond Cutter
- Joined
- Sep 3, 2000
- Messages
- 6,695
It's in stone: All diamonds that look good are not cut good. All diamonds that are cut good do not look good. Cut and Performance need to be separated.....just like a Siamese twins
Bill: The way we grade cut is with separate light behavior and DFS grades. I agree that it takes both sufficient craftsmanship and excellent light performance to define what's best. Consumers are reasonably able to select a diamond with high light performance, but what they cannot generally see is the craftsmanship elements that may also be of long term importance, too.
I think that if we want a system which actually works, we need to get the face-up position of the diamond handled properly. Since diamonds tend to be cut very much alike, I don't think the tilted issue is something that needs to be graded. It follows that a diamond that performs to a high level face up will be among the best at 10 degrees tilt too. Then again, who is going to grade diamonds tilted away from their face-up view? Its a needless exercise and will make grading diamonds much more difficult. What we need is a useful, logical standard.
Bill: The way we grade cut is with separate light behavior and DFS grades. I agree that it takes both sufficient craftsmanship and excellent light performance to define what's best. Consumers are reasonably able to select a diamond with high light performance, but what they cannot generally see is the craftsmanship elements that may also be of long term importance, too.
I think that if we want a system which actually works, we need to get the face-up position of the diamond handled properly. Since diamonds tend to be cut very much alike, I don't think the tilted issue is something that needs to be graded. It follows that a diamond that performs to a high level face up will be among the best at 10 degrees tilt too. Then again, who is going to grade diamonds tilted away from their face-up view? Its a needless exercise and will make grading diamonds much more difficult. What we need is a useful, logical standard.
RE: The way we grade cut is with separate light behavior and DFS grades.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
I''m sorry Dave, but I don''t understand this statement. Would you please explain it to me?
Bill
"The best theories are laid to rest by one insignificant fact".
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
I''m sorry Dave, but I don''t understand this statement. Would you please explain it to me?
Bill
"The best theories are laid to rest by one insignificant fact".
- Joined
- Sep 3, 2000
- Messages
- 6,695
Bill Bray, mayself and many others agree that grading of the best cut in diamonds requires two separate and distinct measures. The first measure is of Light Performance and the second measure is one that serves to define craftsmanship. The DFS grading which we do is fouind on www.gemappraisers.com and in some small way correlates with the AGA Cut Class grades of the past, but is used in a way to show a diamond has sufficient durability, good Finish (polish and symmetry) and also has a reasonable Size appearance, spread to depth relationship. We report light and DFS as separate grades. It takes the best of both to make the very best cut diamonds.
If one combined the two, it is conceivable we''d never find some unsual diamonds which look wonderful, yet have certain physical cut problems such as too shallow depth for great durability, overly deep ones which just look way small for their weight, or some other intersting and pretty combination of parameters where light looks great yet a certain craftsmanship issue may exist. Such stanges diamonds may prove excellent values in earrings, or as small enhancement stones.
Also, there are many diamonds which are reasonably well cut, yet perform only moderately well. Just using physical parameters screens for those which are not well crafted, but does nothing to eliminate the well cut, moderate performer from being overgraded or overpriced. The wide range of the old AGS 0 round is a perfect example of many diamonds being given 0 cut yet not really up to being the best because they were really too deep.
I hope this clears up your questions.
If one combined the two, it is conceivable we''d never find some unsual diamonds which look wonderful, yet have certain physical cut problems such as too shallow depth for great durability, overly deep ones which just look way small for their weight, or some other intersting and pretty combination of parameters where light looks great yet a certain craftsmanship issue may exist. Such stanges diamonds may prove excellent values in earrings, or as small enhancement stones.
Also, there are many diamonds which are reasonably well cut, yet perform only moderately well. Just using physical parameters screens for those which are not well crafted, but does nothing to eliminate the well cut, moderate performer from being overgraded or overpriced. The wide range of the old AGS 0 round is a perfect example of many diamonds being given 0 cut yet not really up to being the best because they were really too deep.
I hope this clears up your questions.
Thank you Dave for explainin that. I agree with you 100% on certain points. The points about grading that I disagree with are these:
1. Physical cutting parameters are much more accurate in designating differences between stones than performance observations as long as the proper measurements are being taken, not averaged or unduly penalized.
2. Finish and durablility should be listed on a grading report but not within a cut anaylysis score/grade. The reason is that finish is sometimes out of the control of the cutter one, and two, if finish is concerned with the absence or presence of polishing lines, then you may have "performance" irregularities because the finish on a facet is more about attaining the best adamatine lustre rather than the absence of lines.
3. Durablility. I''ve stated this before in this forum. Two things about durablility. It''s a saleability factor more than a pricing factor. Why? Sharp edges/points are more and less significant depending upon the utilization of the diamond. Also, durability can''t be cherry picked i.e. take off for knife edge on one spot of a round brilliant''s girdle and not for a corner on a princess cut or a "no culet" on a premium cut stone.
Bill Bray
Diamond Cutter
1. Physical cutting parameters are much more accurate in designating differences between stones than performance observations as long as the proper measurements are being taken, not averaged or unduly penalized.
2. Finish and durablility should be listed on a grading report but not within a cut anaylysis score/grade. The reason is that finish is sometimes out of the control of the cutter one, and two, if finish is concerned with the absence or presence of polishing lines, then you may have "performance" irregularities because the finish on a facet is more about attaining the best adamatine lustre rather than the absence of lines.
3. Durablility. I''ve stated this before in this forum. Two things about durablility. It''s a saleability factor more than a pricing factor. Why? Sharp edges/points are more and less significant depending upon the utilization of the diamond. Also, durability can''t be cherry picked i.e. take off for knife edge on one spot of a round brilliant''s girdle and not for a corner on a princess cut or a "no culet" on a premium cut stone.
Bill Bray
Diamond Cutter
- Joined
- Sep 3, 2000
- Messages
- 6,695
1. Physical cutting parameters are much more accurate in designating differences between stones than performance observations as long as the proper measurements are being taken, not averaged or unduly penalized.
I can live with your statement about differences, but the devices like Sarin/Ogi are full of machine error, averaging, rounding, etc. Nothing is more accurate than a direct measure. A calculation based on corrupted data is pretty much a corrupted final product.
2. Finish and durablility should be listed on a grading report but not within a cut anaylysis score/grade. The reason is that finish is sometimes out of the control of the cutter one, and two, if finish is concerned with the absence or presence of polishing lines, then you may have "performance" irregularities because the finish on a facet is more about attaining the best adamatine lustre rather than the absence of lines.
Finish and durability are addressed by us separately from analsysis of light behavior. Theyare reported separately, too. If we have performance irregularities for any reason, this shows in light analysis. So long as we don''t neglect important data which alters the looks and suitability of a diamond, we have covered the bases. Where it is reported makes less difference than not reporting it or hiding it.
3. Durablility. I''ve stated this before in this forum. Two things about durablility. It''s a saleability factor more than a pricing factor. Why? Sharp edges/points are more and less significant depending upon the utilization of the diamond. Also, durability can''t be cherry picked i.e. take off for knife edge on one spot of a round brilliant''s girdle and not for a corner on a princess cut or a "no culet" on a premium cut stone.
You are right on this. We deduct for any zone on a diamond that falls below slightly thin. This covers thin corner points, or zones on the girlde of a round diamond. The "no culet" is a style issue with stones set rather high today, the no culet comment rarely leads to problems. I have to go with the AGS approach here and let the no culet issue be tabled for the moment. We both recognize it is a better thing to have a small culet, but we are in a minority here.
I can live with your statement about differences, but the devices like Sarin/Ogi are full of machine error, averaging, rounding, etc. Nothing is more accurate than a direct measure. A calculation based on corrupted data is pretty much a corrupted final product.
2. Finish and durablility should be listed on a grading report but not within a cut anaylysis score/grade. The reason is that finish is sometimes out of the control of the cutter one, and two, if finish is concerned with the absence or presence of polishing lines, then you may have "performance" irregularities because the finish on a facet is more about attaining the best adamatine lustre rather than the absence of lines.
Finish and durability are addressed by us separately from analsysis of light behavior. Theyare reported separately, too. If we have performance irregularities for any reason, this shows in light analysis. So long as we don''t neglect important data which alters the looks and suitability of a diamond, we have covered the bases. Where it is reported makes less difference than not reporting it or hiding it.
3. Durablility. I''ve stated this before in this forum. Two things about durablility. It''s a saleability factor more than a pricing factor. Why? Sharp edges/points are more and less significant depending upon the utilization of the diamond. Also, durability can''t be cherry picked i.e. take off for knife edge on one spot of a round brilliant''s girdle and not for a corner on a princess cut or a "no culet" on a premium cut stone.
You are right on this. We deduct for any zone on a diamond that falls below slightly thin. This covers thin corner points, or zones on the girlde of a round diamond. The "no culet" is a style issue with stones set rather high today, the no culet comment rarely leads to problems. I have to go with the AGS approach here and let the no culet issue be tabled for the moment. We both recognize it is a better thing to have a small culet, but we are in a minority here.
- Status
Did You Miss the April 2024 Jewels of the Weeks?
Did You Miss the April 2024 Jewels of the Weeks? - 04/26
Did You Miss The Throwback Thursdays For April 2024?
Did You Miss The Throwback Thursdays For April 2024? - 04/25
300x240.png)