2006 GIA grading report - Post info here please

[JohnQuixote]

Date: 1/25/2006 2:48:14 PM
Author: belle


dave..knowing that is helpful. i have come to understand accuracy is a holy grail. my expectations might have been raised because of some helium hype here

...of course the reality is that i don’t have the same mission as the scientists, i just want help predicting that a diamond may be a good performer for an average buyer. for my purposes sarin reports have proven reliable and helium will too. all of this is just more proof that to judge a diamond *decisively* you must see it with your own eyes..for accuracy and especially for taste. no disrespect to sarin or helium.

could you explain why it is so hard to measure the table?

Belle, in these non-contact measuring devices the diamond is upside down, resting on its table as it is scanned. As Serg mentioned, even a piece of dust on the platform near the diamond may pose a problem.

On PS our accuracy standards are very high, especially for the tech and science people. You’re correct about needing to see a diamond with your own eyes to judge matters of taste, but Helium will be great for broad predictions on rounds, and the fancy bugs will eventually be worked out I suspect.

 

[JohnQuixote]

Date: 1/25/2006 3:54:27 PM
Author: adamasgem

Dave.. You need a combinaton of both methodologies, modeling and direct measurement, in my opinion. This is such that you don't get into the quagmire of basing an overall conclusion on one vewpoint or envrironment for the stone, either of which that the consumer may never see.

Exactly right Marty. As systems become more complex the ‘viewpoints and environments’ do too.

Here’s my simplified view:

> Basic proportions assessment (known ‘proven’ proportions) is simple and consistent within a useful range.

> Basic performance assessment (IS/ASET - standardized) is also consistent within a useful range, and somewhat user-friendly.

Beyond that, it’s not so simple.

> Intermediate proportions assessment includes tools like HCA and Facetware. They are useful but not decisive: Human interpretation plays a role in their metrics so the viewpoint and environment complexities Marty alluded to come into play.

> Advanced proportions assessment is using a 3D scan for software simulation. At this level of complexity human programmers and calculations are involved so the results are subject not only to viewpoint/environment, but also additional complexities like hardware calibration, scan accuracy (being discussed) and software trial/revision.

> Advanced performance assessment includes devices like BS, Imagem & Isee2. Again, escalating complexity means more human interpretation, especially with regard to environment. This is obvious from the fact that results from these machines are different, depending on the company that designed them.


Those who have an interest in a specific method will defend it (or them). The fact is, they are all predictions.

I believe we can meet on common ground that says no system is perfect except the one where the diamond is in-hand. Nothing beats live assessment.

Of course, we are involved on the internet, so live assessment often happens for people AFTER the diamond has been assessed from afar. My opinion – as it relates to internet diamond buying and the consumer – is that a basic level of proportions assessment AND performance assessment for each diamond is a minimum requirement... Beyond that there is a nice buffet of options that continues to evolve.

 

[Garry H (Cut Nut)]

Just a note about proportion grading and performance grading.

Perormance is what we would class as Beauty Grading - the things that differentiate a facetedand polished diamond from a piece of rough.

Dave''s direct assessment means that issues like polish are covered. 3d modelling needs a lab to make a polish assessment. Symmetry effects on Beauty is easily done by both methods.
We would hope that all labs would give similar grades to similar stones of any and all cut / facet structures eventually.
AGS may use a stricter approach - but then the market can and does adjust to a premium for AGS 0 over GIA Ex.

Proportions: Durability, spread and Bill Brae''s artistic / craftsmanship type factors should be expected to also be variable based on different labs standards. For instance it might be expected that labs like GIA, EGL and IGI might have fewer and less strict grade distinctions than ''boutique'' labs.

But what should be most important is that we do not have big differences that we can currently expect where GIA Very Good or Good could be AGS 0, or AGS 4 could be GIA Excellent.

That will be very bad for consumer confidence.

 

[strmrdr]

here is one example that we discussed in another thread of scanner data and DC models vs the actual images:

edit: i cant find the arrows image but they were much closer to the actual than the hearts for the sarin data but as John will be quick to point out the hearts are more important.

 

[Garry H (Cut Nut)]

Date: 1/25/2006 8:08:26 PM
Author: strmrdr
here is one example that we discussed in another thread of scanner data and DC models vs the actual images:

edit: i cant find the arrows image but they were much closer to the actual than the hearts for the sarin data but as John will be quick to point out the hearts are more important.

Excelent comparison Storm.
Dave would you care to make an stl file from Imagem of a hearts stone with a real photo?

 

[oldminer]

Garry, an "stl file" from a "real photo"???? I think you just whizzed right by me. These are things I don''t believe I have at my disposal or I don''t call them by these names.

Give me an email of what you''d like to arrange and I''ll see if ImaGem can supply it on a stone of your preference.

 

[Serg]

Re: princess 6587308„£ags„£73.1, 68.4, 39.7, 37.0, 59.0, 11.8„£gog„£72.7, 66.4, 39.69, 36.78, 59.2, 11.6
princess 6624602„£ags„£72.0, 69.0, 39.6, 37.9, 58.2, 10.9„£gog„£ 71.7, 68.0, 39.55, 37.54, 58.0, 11.0



the depth and tables are far off„£3 of the 4 angle averages on those 2 stones match up like you said in degrees but pavilion and crown % are off.



About direct measurement:

AGS report : 0,826 ct; Height =3,73 Dmin=5,11; Dmax=5,26

Helium : 0,826 ct; Height =3,726 Dmin=5,122; Dmax=5,294

Without Helium file or diamond I can not give great comments, But See mass. If you will decreasing diameters for corresponding AGS report mass will 0,818( It is not proof what AGS report is not correct)

About total heights in %:

AGS report: 73,1%
Helium: 72,75


But
1) 3,73/5,11*100%= 72,99% It is not 73,1%
2) 3,73( roundness from 3,726) /5,12*100=72,85%

About Angle.

See pictures on AGS report, On top picture you can see one Princess cut( with corner facet on crown)
On down pictures You can other Princess cut( without corner facet on crown)

It is NOT important point at all,

About table,

What is methodic of AGS( or SARIN) calculation Table%. Do they use 4 or 2 directions?
We use 2 ( first for min diameter , second for maximum.) In each direction we use corresponding diamond diameter in this direction for normalization.
Id sarin use 4 direction or Sarin use only min diameter for normalization they can receive much bigger table size in %. I do not know Sarin methodic



Sorry for "some helium hype here"

 

[belle]

sergey,
i meant no disrespect to you or helium whatsoever.
i was referring to the fact that people here have helped consumers predict diamonds using measurements for years and sarin was reliable for that purpose. now they find it necessary to go out of their way to criticize it. these are the same people who trusted it while it was the most accurate device available and made many recommendations to consumers based on it...now it''s not good enough.

i am excited about helium (especially when it will work at its best for princess cuts)

i just don’t think its equitable for those to promote helium by insulting another technology they used themselves for years.

you are not doing that. i have great respect for your work and hope i did not offend you.

 

[Serg]

Fixed misprint in previous post. Sorry.

It is NOT important point at all,

 

[adamasgem]

Date: 1/26/2006 12:39:50 PM
Author: belle



sergey,
i meant no disrespect to you or helium whatsoever.
i was referring to the fact that people here have helped consumers predict diamonds using measurements for years and sarin was reliable for that purpose. now they find it necessary to go out of their way to criticize it. these are the same people who trusted it while it was the most accurate device available and made many recommendations to consumers based on it...now it's not good enough.

i am excited about helium (especially when it will work at its best for princess cuts)

i just don’t think its equitable for those to promote helium by insulting another technology they used themselves for years.

you are not doing that. i have great respect for your work and hope i did not offend you.

Belle .. You might be talking about the criticisms I have made

Sarin was the top of the line and leader in "measurement" of diamond, certainly more accurate than the optical profiler we used to use, or microscope reticle, but even then Sarin used to qute angular accuracy of +/-0.2 degrees (but didn't say whether that was a one two or three sigma number). Now they say +/-0.2 degrees, but I still don't have a statistical answer for their "accuracy", one sigma, two sigma, etc or if they mean a confidence bound wth some level (90%, 95%, 98%) of confidence that the "truth" lies between the mean plus or minus some number, all related to the standard deviation about the mean value.

a 90% confidence bound is approxmately +/- 1.86*sigma
a 95% confidence bound is approximately +/- 2.306* sigma
a 98% confidence bound is approximately +/- 2.896* sigma

So we don't know what the published "accuracy" really means, and whether it relates to absolute accuracy or repeatability which are two entirely different things. You can have an entirely repeatable measurement, but the absolute accurate could be WAY off.

People have just started to look at these questions, and have raised criticisms of apparent inaccuracies, as techniques such as raytracing have been evolving, and a finer look has been taken at what makes diamonds tick, and "metrics" are starting to be published, but unfortunately, no confidence bounds on the numbers given.

Statistically, I have criticized GIA for their rounding scheme, because if one looks at it, and the roundng methodology they employ, they are adding statistical uncertainty to the number, and if they don't take into account the uncertainty of the number they are initially rounding, for example, crown angle, and then base a result on that, you could be kidding yourself.

For example, to toot my own horn a little bit, I was the first to look at, or at least publish, the statistical uncertainty in color grading based on data of "best estimates" versus "nominal" positioning of my diiamond master stones, based solely on data given (reluctantly) to me by GIA, at the request of AGS.

I applied this "knowledge base" to my SAS2000 data presentation, and based on the actual data added the variances (one sigma uncertainties squares) that I could compute based on known information, and then added to the uncertainty for "unknowns", a lowest level of basic uncertainy of 20% of a color grade range, for reasons like the fact that GIA "best estimate" of a color grade is quantized to 10% of a color grade, and the statistical difference between the nominal position and their best estimate was on the order of 16% of a color grade range.

There is no FUDGING of the value, there should be a simple statement of a confidence range for the truth, whether it is about color grading or CUT grading.

A REALLY BAD example for this for color grading is shown on a stone whose spectrum was distorted by a colored included garnet crystal, giving a total one sigma estamate of 0.158 (~32%), where one color grade range is 0.5 (per the AGS numerical system), where the typical uncertainty is 0.11 (22%) or very close the the MINIMUM 20% uncertainty I would quote based on the "yardsticks" I have been given as "blessed" by GIA.

 

[JohnQuixote]

Date: 1/25/2006 8:28:59 PM
Author: Garry H (Cut Nut)

Dave would you care to make an stl file from Imagem of a hearts stone with a real photo?

With respect guys, I don’t know that it will do much good to pile on an Imagem vs Helium/DC debate here. There are other threads that could be bumped.

 

[JohnQuixote]

Date: 1/26/2006 12:39:50 PM
Author: belle

sergey,
i meant no disrespect to you or helium whatsoever.
i was referring to the fact that people here have helped consumers predict diamonds using measurements for years and sarin was reliable for that purpose. now they find it necessary to go out of their way to criticize it. these are the same people who trusted it while it was the most accurate device available and made many recommendations to consumers based on it...now it's not good enough.

i am excited about helium (especially when it will work at its best for princess cuts)

i just don’t think its equitable for those to promote helium by insulting another technology they used themselves for years.

you are not doing that. i have great respect for your work and hope i did not offend you.

I think some of this perception is because no one from Sarin posts on the forum. Helium gets its share of promotion & defense and Sarin is criticized, often by the people supporting helium. Thus, there is a perception of one-sidedness that’s not the fault of Serg/OctoNus.

To Belle, putting it in perspective, we’re in a good place now. In 2000 HCA and Ideal-Scope were not available. Ogi/Sarin were far less accurate. Online diamond sales have gained credibility as we are able to better communicate and offer ‘proof’ of performance sight-unseen. Sarin is a great device and Helium will raise the bar for everyone. It is a step forward for accuracy, especially where scientists and analysts are concerned. DiamCalc has helped me understand much about diamonds and I use it daily.

I do understand Belle’s point about ‘discarding’ what has served us for years. It’s not right to suddenly say Sarin isn’t ‘good enough’ for the broad purposes many of us use it for.

The criticisms here – whether of Sarin, Helium, Imagem, BS, or Angelina Jolie

– help the people behind the technology understand what needs to be improved or explained. We are fortunate to have Serg available to answer questions about Helium & DC. Having Marty’s input lately is a great boon to us as well. It’s nice when real scientists stand up and offer objective input.

 

[strmrdr]

helium vs sarin
old vs new
if things didnt push forward we would all be driving wagons and the net would not exist.
technology is changing at an insane pace.

Iv never been happy with sarin data for 3d modeling the models arent close enough.
That the helium data is closer makes me very happy :}
If everything is right in the world it will push sarin to come out with an even better scanner and then helium2 and maybe even ogi will get in the game and we win with more accurate data.
To accept yesterdays technology today is not progress.

 

[Garry H (Cut Nut)]

Date: 1/26/2006 2:02:49 PM
Author: JohnQuixote

Date: 1/25/2006 8:28:59 PM
Author: Garry H (Cut Nut)

Dave would you care to make an stl file from Imagem of a hearts stone with a real photo? Dave an .stl file is a standard 3D file format that your IT guys would know how to make. I believe DiamCalc can open these, or they can easily be converted to a file format that can. Then all we need is an ideal-scope or H&A''s image of the same stone.

With respect guys, I don’t know that it will do much good to pile on an Imagem vs Helium/DC debate here. There are other threads that could be bumped. Good advice John, but the reason I asked was that Dave mentioned the innacuracy of table dimensions when calculated using shadow profiling.
You probably saw in Vegas or maybe Philly - that Imagem use an actual photo in reflected light of the table to measure the table size. This is a very clever idea, and, as Prof Agwarral has made claims of enhanced accuracy of their scanner that takes measurements from diferent orientations - I thought it would be a good and relevant test given the turn of topic here.

Re defending Sarin - I was an active defender of Sarin pavilion angles for many years prior to Helium. Sarin''s method gets its most accurate data on pavilion angles, which per chance are about 10 times more important than table sizes, 5x more important than crown angles etc for actual cut grading importance with hCA.
Checking many Ogi reports on GOG we can see that there is normally about .2 degree lower angles reported than on the GIA data. Sarin gets its best data from very long facet angles - pavilion facets are long. Its weakness is azimuth data - which is important especially for minor facets and building 3D models - in this area sarin appear to ''massage'' the data and make models that do not represent the true recorded data from the scan.
(AGS have gotten round this to some extent by using the raw data from sarin - not what we get from its regular output).

 

[JohnQuixote]

Date: 1/26/2006 10:15:09 PM
Author: Garry H (Cut Nut)

Good advice John, but the reason I asked was that Dave mentioned the innacuracy of table dimensions when calculated using shadow profiling. You probably saw in Vegas or maybe Philly - that Imagem use an actual photo in reflected light of the table to measure the table size. This is a very clever idea, and, as Prof Agwarral has made claims of enhanced accuracy of their scanner that takes measurements from diferent orientations - I thought it would be a good and relevant test given the turn of topic here.

Ah. Not a drinking contest...a recipe exchange. Very good (though I'd still fancy a new thread for it).

 

[adamasgem]

Date: 1/26/2006 10:15:09 PM
Author: Garry H (Cut Nut)

Re defending Sarin - I was an active defender of Sarin pavilion angles for many years prior to Helium. Sarin's method gets its most accurate data on pavilion angles, which per chance are about 10 times more important than table sizes, 5x more important than crown angles etc for actual cut grading importance with hCA.
Checking many Ogi reports on GOG we can see that there is normally about .2 degree lower angles reported than on the GIA data. Sarin gets its best data from very long facet angles - pavilion facets are long. Its weakness is azimuth data - which is important especially for minor facets and building 3D models - in this area sarin appear to 'massage' the data and make models that do not represent the true recorded data from the scan.
(AGS have gotten round this to some extent by using the raw data from sarin - not what we get from its regular output).

I believe what Garry really means is that AGS uses the Sarin generated mesh data (wire frame model) of the facets and not the RAW data that Sarin uses to generate those.

I don't thnk anyone has access to that data.. That is what I'd like to play with to create a simultaneous soluton to the solid model, a methodology I DO NOT BELIEVE anyone is employing. The error in each facet definition is correlated to each and every other facet in the solid as it is a closed form problem. I believe it is not an easy job to do, but it would improve the overall accuracy.

 

[Garry H (Cut Nut)]

John and Marty - You are both right.

(please note - it is friday evening - post japanese restaraunt - and a drop of vino and sake - and I am still nice)

 

[adamasgem]

Date: 1/27/2006 5:42:43 AM
Author: Garry H (Cut Nut)
John and Marty - You are both right.

(please note - it is friday evening - post japanese restaraunt - and a drop of vino and sake - and I am still nice)

So Maybe you need more Sake, more often

Sorry Garry, couldn''t resist, as I didn''t have any Sake

 

[Serg]

Belle,

Last several years I strongly recommended to all our clients of HR and Pacor to use processors Intel. Especially I increasing my recommendations after appearing of Hyperthreading in processors.
Last year I didn''t so insist and even sometimes provide the systems on the base of processors AMD.
At present moment after appearing of processors AMD dual core I recommend to use AMD processors. They are more fast and less noisy.
See... http://www.octonus.com/oct/products/pacor/performance.phtml
I don''t sell any processors. Do you consider that my actions are "some AMD hype here"?

If no, then why is somebody under your criticism if he recommended Sarin early but now he recommends Helium.
It is not attack, I want show to you other side of coin only.

Along with this I am not glad at advertising of Helium on the PS. It is only problem for us. Unfortunately we haven''t opportunity to work on American market for experts.
All our efforts are directed in the Indian market. There is question what technology will be winner and will be more spread.

 

[adamasgem]

Somehow this data magically appeared, someone took the time to play with software
....... AGS 0 vs GIA Ideal Boundaries

I can't take credit for anything other than reformatting to a pdf file

ADDED: I might also add that I believe a GIA EX for cut grade can be gotten with a VG/VG for polish and symmetry where AGS requires an Ideal for both..

I have long felt that way too much emphasis has been placed on polish, where on tiny facet that doesn't take the best polish, probably becuase of the orientaion or possibly a small "burn" from the dop, shouldn't down grade the cut grade, and I think the current AGS stand on polish should be relaxed a bit, so in a way I "agree" with GIA that a VG polish should still be OK for an EX cut grade....

Now as too symmetry, GIA is outright WRONG allowing a VG

 

[Garry H (Cut Nut)]

Is that the CIA, or the KGB, knocking at your door Marty?

 

[strmrdr]

Date: 1/28/2006 12:30:04 AM
Author: adamasgem
Somehow this data magically appeared, someone took the time to play with software

....... AGS 0 vs GIA Ideal Boundaries


I can''t take credit for anything other than reformatting to a pdf file

way kewl
thanks
hmmmm whats most interesting to me is what combos they argree on.

 

[adamasgem]

Date: 1/28/2006 12:54:59 AM
Author: Garry H (Cut Nut)
Is that the CIA, or the KGB, knocking at your door Marty?

Well Garry, since I''m in the CIA database anyhow, and had done some work with and for them in my past, and even had one PhD from the agency following me around for a few weeks writing down everythng I was doing in weapons systems analysis, as a learnng experiance, I''m not worried in the least.

 

[RockDoc]

As the saying goes.................


" It isn''t nice to fool with Mother Nature."


In this case ... it ain''t nice to fool with Marty either!


Rockdoc
(Sitting here in the peanut gallery, drinking Beetlejuice)

 

[JohnQuixote]

Date: 1/28/2006 12:30:04 AM
Author: adamasgem
Somehow this data magically appeared, someone took the time to play with software
....... AGS 0 vs GIA Ideal Boundaries

Off the record: ''Wicked cool Marty!''

On the record: ''Huh? What?''

 

[valeria101]

Date: 1/27/2006 2:54:34 AM
Author: adamasgem

I believe what Garry really means is that AGS uses the Sarin generated mesh data (wire frame model) of the facets and not the RAW data that Sarin uses to generate those.

I don''t thnk anyone has access to that data..

Wait a minute, if not... where do these funky drawings come from?

(someone cited these from Jonathan''s site a while ago and there was some discussion).

 

[strmrdr]

val,
perfect example of a sarin scanner not being able to track facets.
That is why I dont like sarin scans as the bases of a cut grade.
helium tracks them much better and thats not an ad but the simple truth.
Id hope that AGS''s software would kick out an error being fed data like that and the helium machines would be used.
Then the question becomes does the more accurate scan give them an unfair advantage over other diamonds when it comes to grading.

GIA got around the problem by knocking them out of the top grade.

 

[Regular Guy]

The idea wasn't popular when I first suggested it, but...

Date: 1/28/2006 12:30:04 AM
Author: adamasgem
I can't take credit for anything other than reformatting to a pdf file

...with this sort of data, who can resist Ven 0 (the place where AGS and GIA agree on ideal).

 

[JohnQuixote]

Date: 1/28/2006 9:32:55 AM
Author: strmrdr

Id hope that AGS's software would kick out an error being fed data like that and the helium machines would be used.
Then the question becomes does the more accurate scan give them an unfair advantage over other diamonds when it comes to grading.

No worries, because scans don’t have that much say. At AGS anytime a machine judges something that equates to human perception a human is required to confirm it. There are things like girdle plane and culet that only a computer can perform, but when any scientific valuation is produced a human must look at the stone before the assessment is made - including the light performance grade.

This is why only proprietary grades may be distance-given, and why diamonds with painted girdles or similar additive brillianteering can be ‘Ideal:’ Humans confirm the Ideal category of performance.

GIA got around the problem by knocking them out of the top grade.

I’m still not aware of what level of non-traditional brillianteering GIA considers ‘significant,’ but if they are dismissing any painted/dug girdles out of hand it’s another major gaffe that needs to be thrown on the fire along with rounding.

 

[valeria101]

Date: 1/28/2006 9:32:55 AM
Author: strmrdr
val,
perfect example of a sarin scanner

I know... that was a post about access to data. It looks like the charts on his site use the type of data called upon in Mart Haske''s post

Btw. I don''t know precisely how much less correction Helium readings need to add up to a realistic graphic representation. As long as you guys decide it is a revolution, it is good enough for me.