GIA has started providing the cut grade dimensions

[Garry H (Cut Nut)]

That is a reasonably honest match Marty. I have seen examples where the software try''s to average out the errors.

This is an example from the sarin website.

 

[Serg]

re:Here is a Sarin Mesh file that a friend of mine who has a $20$ Sarin did on an Eightstar. I would say that the representation of meet pont facetining is a little screwed up, and based on that, I wouldn't trust the facet angles at all. I decoded the srn file and pulled out the Mesh data on the left and the right hand side was the Sarin webviewer. Makes you think that something is wrong, doesn't it, and IT ISN'T THE DIAMOND..

*8 diamond has very small degree between main crown and girdle facet( 3-5 degree) . Classical round has 11.25 degree.

If you have a very small error between two such( 3-5 degree) facets it could give big mistake for meet point.

Such stones is problem for any shadow scanners( And for Helium of course too ) Helium has less error than Sarin , but has.

Such error is not problem for calculate light performance if angles of facets are correct.

 

[adamasgem]

Date: 11/28/2005 5:13:23 PM
Author: Garry H (Cut Nut)
That is a reasonably honest match Marty. I have seen examples where the software try''s to average out the errors.

This is an example from the sarin website.

Great example of "accuracy", no wonder we have averaging and rounding. But that shouldn''t be the basis for a cut grade system, just because some can''t measure correctly. Looks like we have been putting two much faith in the numbers we are given.. I think Helium has resolved that issue.

As the stone gets smaller the accuracy degrades also for a fixed optics system...
On machine measures differently than another on the same stone, etc etc.. But roundiing just increases the error in errored data.. Didn''t they ever hear of weighted estimates based on spline or linear interpolation methods instead of arbitrarily roundiing the data to fit their database? Geeze, absolute stupidity, in my opinion.

By the way, the stats I showed on AGS grading based on "availability" are not entirely inconsistent with larger sample sets, which may be published later. I''m working on that.

 

[adamasgem]

Date: 11/29/2005 2:33:38 AM
Author: Serg

re:Here is a Sarin Mesh file that a friend of mine who has a $20$ Sarin did on an Eightstar. I would say that the representation of meet pont facetining is a little screwed up, and based on that, I wouldn''t trust the facet angles at all. I decoded the srn file and pulled out the Mesh data on the left and the right hand side was the Sarin webviewer. Makes you think that something is wrong, doesn''t it, and IT ISN''T THE DIAMOND..

*8 diamond has very small degree between main crown and girdle facet( 3-5 degree) . Classical round has 11.25 degree.

If you have a very small error between two such( 3-5 degree) facets it could give big mistake for meet point.

Such stones is problem for any shadow scanners( And for Helium of course too ) Helium has less error than Sarin , but has.

Such error is not problem for calculate light performance if angles of facets are correct.

Same type of pproblem with sample srn files downloaded when I installed the web viewer.. Question is, how can the angles be correct if the mesh is not, or can you rely on them

 

[adamasgem]

Date: 11/27/2005 12:14:19 PM
Author: oldminer
Rounded up numbers mean greater overall innaccuracy in a predictive model.

I agree

The numbers need to be rounded as the equipment used to generate the numbers has fairly wide machine error which has not yet been cured. Someday, they may have better measuring tools in the lab.

But rounding only increases the error

It is just laziness on GIA''s part not wanting to interpolate in their software..

Suppose the original numbers were actually "perfect", then rounding the way they do, screws up the grading.. Besides, shouldn''t a more accurate machine be rewarded with a potentially better overall grade, based on the measuring machine''s estimate of the "truth", and not have GIA mucking it up?

They can always put confidence bounds on the answer