AGS Lab releases VPA software

[Serg]

Peter Yantzer from AGSL gave permission to publish his answer:



"
From: "XXXXXX>
Subject: GIGO
Date: 29 May 2012 03:27:00 GMT+03:00
To: Sergey Sivovolenko , Garry Holloway

Hi Sergey:

I respectfully submit that you are missing the point:
In a short period of time the manufacturers will have to provide accurate scans. If they don't, either wholesalers, retailers or consumers will reject GI (garbage in) scans. This is irrespective of whether or not the stone has a tipped table causing the VPA image to look worse than the stone really is. They won't buy it or will reject it out of hand. Do you think they will want to learn that X stone is not that bad because the table is slightly tilted? I don't think so.

So, to answer question 1: Consumers won't have to know if a scan is GI (garbage in). The scan will have been checked or approved by the distribution chain.
To answer statement 2: You use a tilted table as the cause of increasing garbage out. A tilted table is a major symmetry fault. So if a tilted table causes garbage out and the stone is rejected then so be it! We are still going to do that study I promised Garry about tilted tables and I will report the information to him. Nevertheless, a major symmetry fault is a major symmetry fault. You can do the math. The repair can cost a lot of weight or to minimize weight loss the light performance may be severely lessened because the crown angles have to be reduced, unless you are willing to bring in the diameter at significant loss.

Have you criticized BrillianceScope and other pixel counting devices for the same tilted table problem? Or, are you saying that the light performance is not that adversely affected but the facet angles and facet directions are more adversely affected?

Garry, you tried top make the argument that a piece of dust can get into the press pot before the stone is pressed into it. That argument doesn't fly with me. "xxxxxxxxxxxxxxx"

And, I told you in a previous email that we can adjust a scan to remove the tilted table bias. We just aren't going to do that yet and until the VPA is firmly established in the world.

Feel free to post the above if you like.

Heartily,
Peter






re:Keep in mind that GIGO ( garbage in = garbage out ) definitely applies here.

1)How can consumer define that scan GI is?

2)some software increase garbage other reduce garbage.

for example software that use light performance metrics reduce garbage in ( for example table tilt)
in same any software ,that use directly facet angles with reference to 3D model table , increase garbage out "

P.S
I changed one Peter statement to ""xxxxxxxxxxxxxxx""

 

[Serg]

Hi Peter,
Thanks for possibility to discuss this technical issue on PS.

Please find below my comments:


I respectfully submit that you are missing the point:
In a short period of time the manufacturers will have to provide accurate scans.

Happy to see such Future. How long is short period? 1 year? 5 years?

If they don't, either wholesalers, retailers or consumers will reject GI (garbage in) scans.

How many wholesalers, retailers , consumers do use scans now? Unfortunately pure scans are not important for consumers.
Consumers needs comparison of Light Performance between different scans . Light Performance is more important than Craftsmanship for most consumers.

This is irrespective of whether or not the stone has a tipped table causing the VPA image to look worse than the stone really is. They won't buy it or will reject it out of hand. Do you think they will want to learn that X stone is not that bad because the table is slightly tilted? I don't think so.

Lets separate two cases:
1) Diamond has table Tilt, scan Has similar table tilt( scan has good enough accuracy)
2) Diamond has not Tilt( tilt is less than 0.1 degree for example), scan has big tilt( 0.5 degree or more).
I suppose what diamond (ascetic.74.srn ) has not tilt in 1 degree as scan has.
why consumer have to reject diamond without tilt table if scan has tilt table?


So, to answer question 1: Consumers won't have to know if a scan is GI (garbage in). The scan will have been checked or approved by the distribution chain.

Do you mean Checked, Corrected and Approved?
If scans become important for retailers business then consumer will see mainly corrected ( manually of automatically) 3D models.
Many OctoNus Indian clients asked me to develop algorithm which increase model quality after scan. Usually Diamonds have much better minor and major symmetry than scans . cutters are not happy use scans which penalty Excellent cuts. We can develop such software but we did not it . I prefer to build bad model for good diamond than good model for bad cut.
If Helium scan model has very good minor symmetry then consumer can be sure what diamond has not worse minor symmetry than scan model.

To answer statement 2: You use a tilted table as the cause of increasing garbage out.

I considered Model table tilted as Garbage In.( I suppose what this particular diamond has not so big table tilt)

A tilted table is a major symmetry fault.

I am disagree if we speak about Optical Performance. Table tilt is just most easiest detected by Labs( scanners) Craftsmanship fault. Major Craftmanship fault is not specially Major Optical Performance fault.
there are many other more critical cut deviations for Optical Performance

So if a tilted table causes garbage out and the stone is rejected then so be it!

have we reject diamonds with 0.2 degree table tilt? 0.4 table tilt? where is boundary ?


We are still going to do that study I promised Garry about tilted tables and I will report the information to him. Nevertheless, a major symmetry fault is a major symmetry fault.
You can do the math.

I did math. 0.2 degree Pavilion facet deviations give much more impact to Optical performance than 0.2 degree table tilt.
false 0.2 degree Table tilt in scan gives false 0.4 deviation for crown and pavilion angles

The repair can cost a lot of weight or to minimize weight loss the light performance may be severely lessened because the crown angles have to be reduced, unless you are willing to bring in the diameter at significant loss.

I sure there are no reason to repeat diamond if just scan model has table tilt.

Have you criticized BrillianceScope and other pixel counting devices for the same tilted table problem?


BrillianceScope has not same tilt table problem. Brilliancescope has many other problems( I published it on PS may years ago)
there are two Light Performance technologies ( direct and indirect). Indirect Light Performance ( Based on scans) can partially compensate Garbage In ( such as errors in table tilt, additional facets, etc)


Or, are you saying that the light performance is not that adversely affected but the facet angles and facet directions are more adversely affected?

Yes, Integral Light performance calculations are less sensitive to scan errors than statements based on pure facets angles


Garry, you tried top make the argument that a piece of dust can get into the press pot before the stone is pressed into it.


I think scan table tilt come mainly from bad scanner holder alignment .

That argument doesn't fly with me. You can eat off the floor in Indian cutting factories.

And, I told you in a previous email that we can adjust a scan to remove the tilted table bias. We just aren't going to do that yet and until the VPA is firmly established in the world.

Feel free to post the above if you like.


P.S wrong table tilt is not PVA mistake. PVA mistake is wrong facet type recognition . 2 main Pavilion facets had been considered as Pavilion Halves that increases deviation in slope and mainly direction angles in PVA report
You can correct PVA to avoid such mistake. If we speak about Full round cut your developers can use facet width to separate halves from main facets. Halves always have width near girdle when Main facets have width far a way from girdle plane.
Unfortunately such simple criteria does not for for semicut and we can not use it in Diamcalc( may market for Diamcalc are semicuts)

Best Regards

 

[Karl_K]

All from Saran HD and generated from the files above:

Image on the left:
One slightly shallow:


One slightly steep:


one steep one shallow


Bunch off steep and shallow:


Which is it?

 

[Karl_K]

I could go one for hours posting more examples but I have got to hit the hay.

7? years ago I proposed that "tightness" be a criteria when considering which diamonds to buy.
I came to the conclusion that variation between scanners and basic scanner accuracy made that impractical.
I have not seen anything here to change my mind about it.

 

[Garry H (Cut Nut)]

To all, is there a simple way to check a scanner for table flatness / horizontal-ness relative to the axis of rotation?
There is no instrument with moving parts that is perfect. And any inaccuracy there will certainly lead to as doubling of error in crown, pavilion and culet offset.

Peter is it possible to have your calibration sample stone sent for goniometer analysis? Or to GIA's new helium scanners ( what software version are or were you using? It is now up to version 5.4 http://www.octonus.com/oct/download/helium_download.phtml ).

 

[Serg]

Serg can you open sample 13 in DC. Scan error or possible parsing error in DC?
edit: sample 20 looks questionable to me also.

Karl,
most probable sample 20 had "Big Dust" on crown and girdle during scan.
I can not add more information about sample 13 without source information( contours )

 

[Serg]

All from Saran HD and generated from the files above:

Image on the left:
One slightly shallow:


One slightly steep:


one steep one shallow


Bunch off steep and shallow:


Which is it?

Karl, please find VPA report and Diamcalc report for Same model!

VPA does not account Diamond model Tilt. diamond model tilt is not same as table tilt.
scanner holder plane can not be exactly parallel . OctoNus software account tilt of scanner holder plane. Neither modern Sarin scanner nor PVA does not account such holder tilt. You can see what all Sarin scans have zero slope angle for Tables, but all Helium scans have not zero slope angles for tables.
it is main reason why even old 2004,2005 Helium scanners give better concistency than modern Sarin scanners.
FYI AGS bought Helium scanners in 2004,2005 years, we delivery its from Moscow and Never come to USA for any hardware alligment .

just check table facets for Sarin and helium scans. Do you see difference in facet junctiones?

 

[Serg]

Diamcalc report for Sarin scans 2009 by AGS, 3 scanners, same diamond

 

[Serg]

Diamcalc report for Sarin scans 2012 by AGS, 4 scanners, same diamond

 

[Serg]

Diamcalc report for Helium scans 2009 by AGS, 1 scanner, same diamond

 

[Serg]

table junctiones
scan Sarin HD 2012
scan Helium 2009, delivery 2004/2005

same diamond, ASG data

 

[japroffitt1]

I just ran a diamond I was considering and definitely helped me exclude it. The scan seems to have been bad, but the diamond was already shipped backed to my jewelers local office from their main office that has the Sarin so I was unwilling to take a risk on it. Very helpful tool!

Hi japroffitt1,

Could You please send me the address of your jeweler?
I want ask him to scan this diamond on any other scanner .

Serg, He no longer has the diamond in hand. He had called it in at my request and has since returned it. I will be sure to post any future Sarin scan files that I receive as you all are certainly much more adept at analyzing them than I am. Thanks for offering to help.

 

[Karl_K]

table junctiones
scan Sarin HD 2012
scan Helium 2009, delivery 2004/2005

same diamond, ASG data

I noticed that also.
It does not give me faith in using the sarin results to measure craftsmanship.

 

[Karl_K]

To all, is there a simple way to check a scanner for table flatness / horizontal-ness relative to the axis of rotation?
There is no instrument with moving parts that is perfect. And any inaccuracy there will certainly lead to as doubling of error in crown, pavilion and culet offset.

I have no idea, but I wonder if garbage in detection could be programed.
Checked meet points and facet position or something and kick out a bad scan error.
But that would not help with basic scanner accuracy problems.

 

[Serg]

To all, is there a simple way to check a scanner for table flatness / horizontal-ness relative to the axis of rotation?
There is no instrument with moving parts that is perfect. And any inaccuracy there will certainly lead to as doubling of error in crown, pavilion and culet offset.

I have no idea, but I wonder if garbage in detection could be programed.
Checked meet points and facet position or something and kick out a bad scan error.
But that would not help with basic scanner accuracy problems.

Karl,

Just single 3D model is not enough to separate error in model and bad junctions on real diamond. there is necessary additional information.

 

[Texas Leaguer]

As I mentioned we have begun to compare VPA images against diamonds in our stock to see if it can be used as a selection or validation tool for our particular needs. Nothing concrete to report at this time. We are focussing on differences that might be aparent in VPA between our ACA brand and ES category H&A diamonds. Since the differences can be very slight in some cases, it may be difficult to tie in any particular image patterns definitively to one or the other. But we may find some telltale signatures for the ones that are not right on the borderline. We will report anything we can discern in our observations.

Meanwhile, I think it is really interesting to sit on the sidelines and watch this competition of ideas. Healthy debate can bring out issues that need to be further explored and lead the way to possible improvements. At the very least to a better understanding of the intended use and proper interpretation of the results, and the weight to give them in light of other information available about a diamond.

Peter has been clear about the relative nature of the VPA reports, and the fact that they are fully intended to present a graphical view of precision. But Karl has a point, it seems to me, that instrumentation error may not necessarily be consistent from facet to facet which could reduce the reliability of the images. In addition to his comments about the accuracy of various devices and the issue of how the integrity of 3d scans will be vetted and approved in the supply chain, Sergey brings up challenging questions such as the extent to which 3d scans will become fundamental to the consumer, if ever. There are not many jewelers talking to consumers about the kind of things we are discussing here, nor are many manufacturers having discussions with jewelers about them for that matter. And unless GIA should miraculously become interested in making detailed cut information available to the consumer, it seems unlikely to me that it will be become mainstream any time soon. Then again, the pace of change is accelerating, so it may be here sooner than we think!

I see the VPA as a promising new tool to add to our understanding of diamond cut. Whether it becomes a catalyst for major change in the industry as Peter suggests, only time will tell. But it certainly is an innovative way to use information derived from modern measuring devices (in use the world over) to understand specific aspects of diamond craftsmanship. As such it seems logical that it will become very interesting to alot of consumers, at least among the growing portion of the market who understand that cut quality is of key importance. As Peter argues, (paraphrasing) if consumers find something of value they will demand it of their suppliers and that will serve to determine the behaviour of jewelers and manufacturers. At least that's the way it's supposed to work in a free market economy.

 

[Serg]

I think the directional difference between Halves and halves position in classical round is better information for consumers.

for this sample it gives 0.27 degree instead 11.52 degree for Average Lower