describing the painting in report . We ask for more input suggestions…

[adamasgem]

Date: 9/11/2006 10:26:13 PM
Author: canuk-gal
HI:

I acknowledge I am swimming with the big fish here, and hence preface my post by declaring my experience in developing rating scales for research projects, was limited to measuring attitudes.

That said, is it possible to consider using a 7 point linear scale with verbal labels on the end points and the indicators in between labelled with numbers? Diagrams, prn.

Rationale for this suggestion is:

1) 7 points max. based on psychological research by Miller (1956) showing that people have difficulty reliably making more than 7 distinctions (it''s dated, but it''s relevant!). I believe Serg/John mentioned only 7 choices in an earlier post.

2) Research shows that although verbal labels can clarify the meaning of scale points, they may also be distracting. Weisberg et al (1996) surmises it is best to include them, only when necessary. Moreover, if evidence in the literature suggests that people see very little difference between the rating ''Good'' and ''Very Good'', wouldn''t it just be better to use ''the numbers'' and have only the ''extreme'' descriptors/labels on the ends of the scale?

Just an alternative/thought.

cheers--Sharon

As John suggested, a big love note from me

 

[JohnQuixote]

I can feel the love.

I know this is secondary to nomenclature, but I'm still interested in why the varying widths exist in the ranges.

 

[Serg]

(0.0 - 0.1) = 0.1
(0.1 - 0.3) = 0.2
(0.3 - 0. = 0.5
(0.8 - 1.2) = 0.4
(1.2 - 1. = 0.6
(1.8 - 2.2) = 0.4
(2.2 - oo) = 0.8


First interval
1) is necessary due limited accuracy of any current and future scanner.
2) Such painting( or digging out) does not matter for diamond appearance
Second interval
1) Has boundary near 1 degree. Psychological boundary
2) Cover range accuracy most non super symmetrical round diamonds.
3) Most probably such painting( and dig out) is not important for diamond appearance

Fourth and sixth intervals come from current cutters techniques ( 1 and 2 clicks + reasonable range for accuracy) .

Third and fifth are between Fourth and sixth intervals.
Seventh.
1) I do not believe what cutters has possibility and request cut diamond in range 2.6-3.0. Such additional range is not necessary simply.
2) Do you anybody see reason to distinguish 9 and 10 degree painting?)

We can do more equal scale by increasing Fourth and sixth intervals. But I think right size Second( or Second +First) , Fourth and Sixth intervals is more important then equal scale.

Other possibility:

(0.0 - 0.1) = 0.1
(0.1 - 0.2) = 0.1
(0.2 - 0. = 0.6
(0.8 - 1.2) = 0.4
(1.2 - 1. = 0.6
(1.8 - 2.2) = 0.4
(2.2 - oo) = 0.8

 

[Serg]

Re: I just took note of the mesh question and wondered how much the angle errors would be.. Probably the slow scan would improve the "results". Would you have any estimate on the error level in azimuth and tilt for the breaks? It will be worse than the mains I believe because the baselengths are shorter...

Marty,
Length is not most important limitation of accuracy for main and girdle facets.
Accuracy slope angle strongly depends from accuracy azimuth angle .
Azimuth angle accuracy has different type limitation. I want show one reason which does not depends from scanner. Even diamond facets have limited flatness. Take for example 3 microns for crown girdle facet 6 mm round diamond. It give 0.17 degree. Edges have limited sharpness. What is right approximation real diamond by ideal polyhedron?
For symmetrical diamond( with “symmetrical rounded edges and facets”) there are two possibility for position facet ( polyhedron facet keep real edges or at a tangent to real facet)
For non symmetrical facet roundness these two solutions have different facet angles.
You hàve choice between “right angles” and fine facet junctions.( It is reason ‘Bad’ 3d models of CZ samples)

 

[Serg]

re:I do not like maximum because from 7 to 11.75 is a long way.


Garry ,
Tis practical maximum( not theoretical maximum. Sometimes we can not achieve theoretical maximum at all.

 

[adamasgem]

Date: 9/12/2006 1:50:07 AM
Author: Serg

Re: I just took note of the mesh question and wondered how much the angle errors would be.. Probably the slow scan would improve the ''results''. Would you have any estimate on the error level in azimuth and tilt for the breaks? It will be worse than the mains I believe because the baselengths are shorter...

Marty,
Length is not most important limitation of accuracy for main and girdle facets.
Accuracy slope angle strongly depends from accuracy azimuth angle .
Azimuth angle accuracy has different type limitation.
Sergey, I do understand the issues of resolver accuracy (and repeatability) in defining the minimal three points (or two intersecting straight lines) which could define a plane.

With "errored" data, things get messy. I used the baselength example in an attempt to illustrate that the larger the separation between two points (which also define a straight line and a "slope" along the direction of the straight line, but not necessarilt that of the plane), the small the effects of quantization error on the accuracy of the slope.

I want show one reason which does not depends from scanner. Even diamond facets have limited flatness. Take for example 3 microns for crown girdle facet 6 mm round diamond. It give 0.17 degree.
I''m not sure, but aren''t you confusing "flatness" with "roughness" because of the grit used for polish.

Edges have limited sharpness. Yup, that is why the intersection of planes to define facet edges and the intersection of lines (or three planes) define the "virtual" vertices

What is right approximation real diamond by ideal polyhedron? As a start, I would probably use twice the expected number of vertices and let them converge, if in fact they do. This gets messy depending on the x,y,z quantization.


For symmetrical diamond( with “symmetrical rounded edges and facets”) there are two possibility for position facet ( polyhedron facet keep real edges or at a tangent to real facet)
I don''t think there is such an animal possible truely possible..("true" symmetry). Besides one can''t really "measure" it.

For non symmetrical facet roundness these two solutions have different facet angles.
You hàve choice between “right angles” and fine facet junctions.( It is reason ‘Bad’ 3d models of CZ samples) I guess some people don''t understand this.

The analogy is how you have to approximate a bruted girdle but a series of planes, likewise the junction of two planes mat not be a a line, but because of "rounding of the edges" might best be approximated by another plane, and see if the D in the planer equation (Ax + By + Cz +D = 0) goes to zero in the simultaneous solution of the entire closed form. Likewise, each vertex is actually, initially, a "rounded" surface initially approximated by a plane.

Extra facets, cavities, naturals compound the "a-priori" problem. I think you have done remarkably well, even given my knitpicking regarding the meet point faceting issue.

I also think there "may" be methodologies and techniques best discussed offline...

 

[adamasgem]

Date: 9/12/2006 1:11:01 AM
Author: Serg

(0.0 - 0.1) = 0.1
(0.1 - 0.3) = 0.2
(0.3 - 0. = 0.5
(0.8 - 1.2) = 0.4
(1.2 - 1. = 0.6
(1.8 - 2.2) = 0.4
(2.2 - oo) = 0.8
I take it we are discussing the MEAN values of the 16 break facet azimuth shifts from 0
First interval
1) is necessary due limited accuracy of any current and future scanner.
Do you consider this a two sigma inherent bound on the angular accuracy, somewhat like others woud say 0.0 +/- 0.1 ??

2) Such painting( or digging out) does not matter for diamond appearance "Maybe", but if and only if, the std deviation about the mean is very small. It is a generalization that may not be true, if the stone does not have symmetry.

Second interval
1) Has boundary near 1 degree. Psychological boundary Psychological Boundary ???
2) Cover range accuracy most non super symmetrical round diamonds. There probably is a problem with the accuarcy and wearing and reseating of the index wheels.
3) Most probably such painting( and dig out) is not important for diamond appearance

Fourth and sixth intervals come from current cutters techniques ( 1 and 2 clicks + reasonable range for accuracy) . I think I understand you here, you are speaking about the type of painting from Whiteflash (4) and EightStar(6)

Third and fifth are between Fourth and sixth intervals.
Seventh.
1) I do not believe what cutters has possibility and request cut diamond in range 2.6-3.0. Such additional range is not necessary simply. OK
2) Do you anybody see reason to distinguish 9 and 10 degree painting?) NO

We can do more equal scale by increasing Fourth and sixth intervals. But I think right size Second( or Second +First) , Fourth and Sixth intervals is more important then equal scale.

I see whatyou are getting at. Thanks Sergey for explaining your rationale.

Other possibility:

(0.0 - 0.1) = 0.1 Limitations of measurements
(0.1 - 0.2) = 0.1 May be indicative of painting to get good girdle thickness call
(0.2 - 0. = 0.6 Atypically applied
(0.8 - 1.2) = 0.4 Typically applied by one or more cutters
(1.2 - 1. = 0.6 Atypically applied
(1.8 - 2.2) = 0.4 Typically applied by one cutter
(2.2 - oo) = 0.8 Very Unusual

I do think you should state the standard deviation about the mean.

 

[Serg]

re: Do you consider this a two sigma inherent bound on the angular accuracy, somewhat like others woud say 0.0 +/- 0.1

+/- 3sigma for average value of painting even for bad scanner

Even If scanner has +/- 0.5 degree angle accuracy( +/- sigma ), Error is less then 1 degree. Most measurements will have error less than 1.5 degree( 6 sigma inherent bound)
Accuracy for average painting( 16 points) will be in 4 times better . +/-0.125 degree
0.1clicks=0.375
0.125*3=0.375=3 sigma for average value of painting
0.1 clicks range is enough for scanner with sigma=0.5 degree

 

[Serg]

re: I think I understand you here, you are speaking about the type of painting from Whiteflash (4) and EightStar(6). Typically applied by one cutter. Typically applied by one or more cutters


If any cutter uses the classical equipment for round cut, he has choice between 3 Typical possibility
1) 0 click
2) 1click
3) 2 click
Eightstar choice is 2 click
Whiteflash choice is 1 click
But possibility(existence) for such choice does not depend from Eightstar or Whiteflash.

 

[He Scores]

Maybe Paul can help me out here. I''m not aware of any classical cutting equipment that sports "clicks" for variation in either direction or angle for brilliandeering facets. My dops "click" for starting position of direction left or right but that''s about it.

I''m not sayin'' anything, I''m just sayin''. I have old stuff and new stuff but none that meet the click criteria.

Perhaps someone should enlighten me, please?



Bill Bray
Diamond Cutter

 

[adamasgem]

Date: 9/12/2006 11:09:56 AM
Author: He Scores


Maybe Paul can help me out here. I''m not aware of any classical cutting equipment that sports ''clicks'' for variation in either direction or angle for brilliandeering facets. My dops ''click'' for starting position of direction left or right but that''s about it.

I''m not sayin'' anything, I''m just sayin''. I have old stuff and new stuff but none that meet the click criteria.

Perhaps someone should enlighten me, please?



Bill Bray
Diamond Cutter

Any number of click index wheels can be somewhat easily made to fit any dop. I think a lot of the variation is done by using the "cheater"

 

[adamasgem]

Date: 9/12/2006 10:55:43 AM
Author: Serg


re: I think I understand you here, you are speaking about the type of painting from Whiteflash (4) and EightStar(6). Typically applied by one cutter. Typically applied by one or more cutters


If any cutter uses the classical equipment for round cut, he has choice between 3 Typical possibility
1) 0 click
2) 1click
3) 2 click

Eightstar choice is 2 click
Whiteflash choice is 1 click

But possibility(existence) for such choice does not depend from Eightstar or Whiteflash.

Observation One "click verses two. If you maintain the same max mins on girdle thickness (or range) the greater the indexing, the smaller you have to make the separation between the mains. So you have a theoretically greater total weight loss with higher degrees of painting, but higher degrees of painting reduces the dark girdlle edge areas and the stone appears diametrically larger. The distribution of light intensity radially from the center of the stone also changes..

 

[He Scores]

I''m trying hard to follow this Marty, but if the "clicks" are distance AWAY from the main facet, the clicks are still not the sole causation of the g min. It''s virtuallyl impossible to complete a brilliandeering facet with using either one two or three clicks without adjusting the angle adjustment screw.

Pasting, IMHO, is more a product of shallow facet angle, while digging is more a product of facet straightness or "indexing" as you put it.

Pasting will have a greater g min measurement and digging will have a lesser g min that the g bzl %.

Bill

 

[adamasgem]

Date: 9/12/2006 1:10:38 PM
Author: He Scores


I''m trying hard to follow this Marty, but if the ''clicks'' are distance AWAY from the main facet, the clicks are still not the sole causation of the g min. It''s virtuallyl impossible to complete a brilliandeering facet with using either one two or three clicks without adjusting the angle adjustment screw.

Pasting, IMHO, is more a product of shallow facet angle, while digging is more a product of facet straightness or ''indexing'' as you put it.

Pasting will have a greater g min measurement and digging will have a lesser g min that the g bzl %.

Bill

I think the confusion has been caused by the "definition" of click, I think started by Sergey, but I''m not sure. One click off of a typical dop is +/- 11.25 degrees off the main azimuth . Sergey then divided this by 3.75 degrees and called it a click, and not thinking about it labeled my graphs as "click" when they are in fact Increments of 3.75 degrees from the Nominal break facet index position (11.25 degrees from the main)..... Probably the painting offsets are done with the "cheater", but one could have special index wheels made for a particular head to set the "off nominal" break position. Sorry about that if I caused any confusion by not recognizing the nomenclature issue..

 

[adamasgem]

Date: 9/12/2006 1:10:38 PM
Author: He Scores


I''m trying hard to follow this Marty, but if the ''clicks'' are distance AWAY from the main facet, the clicks are still not the sole causation of the g min. It''s virtuallyl impossible to complete a brilliandeering facet with using either one two or three clicks without adjusting the angle adjustment screw.

Pasting, IMHO, is more a product of shallow facet angle, while digging is more a product of facet straightness or ''indexing'' as you put it.

I guess it is a nomenclature issue, as the actual nominal angle is set by the index position and the tilt angle of the dop, as you point out. The "index" we are talking about is the rotation angle of the facet plane normal with about the table /culet Z axis, where the table defines the x-y plane, a perpendicular to that plane defines the Z axis, and a rotation about the Z axis defines the azimuth or index position.

Pasting will have a greater g min measurement and digging will have a lesser g min that the g bzl %.

Bill

 

[Rhino]

How are ya Marty

Thoughts below.

Date: 9/11/2006 3:49:07 AM
Author: adamasgem

Date: 9/10/2006 10:09:05 PM
Author: Rhino


Excellent suggestion John.

Scary how much we think alike at times.

I didn''t like the word ''practical'' as it implies anything other than practical is impractical and moderate IMO is the best word for the job if you want to go through with this Sergey.

Approaching this from a different angle and employing all of the ideas invoked here I would make the following suggestion.

Use the numbers approach Sergey as Marty is suggesting. When you personally put descriptors there even though your intent is not to grade, it will be interpreted that way. I know this becuase I hear it from consumer perspective who consult Helium Reports.

What I would suggest is along the lines of what John has suggested in another post in this thread listing not your descriptors but GIA''s and AGS''s thresholds for their top grades.

If you were going to use descriptors here''s what I''d like to see.

1st term Negligible: Would typically receive AGS0, GIA Ex in cut
2nd term Very Slight: AGS0, GIA EX
3rd term Slight: AGS 0, GIA EX
4th term Moderate: AGS 0, GIA EX (VG at upper limit of scale)
5th term Substantial: AGS 0, GIA VG
6th term Very Substantial: AGS 1, GIA G (VG at lower limit of scale)
7th term Maximum: Lower in both metrics

The ''problem'' with the above (adding AGS and/or GIA) is that is a potential grade effect IF AND ONLY IF other parameters fit the grade range.

I understand your point Marty however all other parameters can be "ideal" or "excellent" yet be penalized by either lab for these features in girdle cutting alone which is why I favor their inclusion. I had forwarded AGS a model of a stone with what most of us here would consider rare and fine craftsmanship in an H&A yet even with ideal angles and precise optical symmetry the painting dinged the stone in both GIA and AGS systems.

Regards,

 

[Serg]

4 step gear wheel 1 click 90 degree
8 step gear wheel 1 click 45 degree
16 step gear wheel 1 click 22.5 degree
32 step gear wheel 1 click 11.25 degree
96 step gear wheel 1 click 3.75 degree

Moscow night time

 

[Rhino]

Date: 9/11/2006 6:43:16 AM
Author: adamasgem

Date: 9/11/2006 1:23:10 AM
Author: Rhino

Excellent point. A correllation I found of interest when you published the newer Helium reports Sergey were your descriptors and how they correllated with GIA''s observation testing results. What you have described as painting up to ''moderate'' makes a GIA Ex (except for perhaps the outer skirts of the moderate range which may get VG, don''t quite know yet). The descriptors you currently have as ''big'' and ''very big'' do not make Ex grade.

I have a few stones here with varying degrees of painting. In the ones I''ve examined I can determine when this feature impacts face up appearance. I''ve also been corellating this with other technologies. Sergey ... I know how I might perhaps help with the descriptors. Do you have .dmc files of stones with the various degrees of notching (minimum and maximum) within each range on your scale? Take the ''moderate'' or ''practical'' range. Do you have handy .dmc files with .8 notches and 1.2 notches? I''d generate them myself but will not be at my DC computer until Tuesday.

Regards,

Rhino.. One of the problems wiith nomenclature deals with perceprtion of goodness or badness.

I understand where you''re coming from Marty but at the same time we should not hide the facts from the public that large, very large or huge painting is in fact large/very large/huge painting regardless of the perception of the public. Take other diamonds with different cutting features that use the same nomenclature. I have clientele who just LOVE old mine cuts. Their culets are described as "large" or "very large". Certainly a negative connotation in the minds of most. Does it stop those people from purchasing old mine cuts? Marty ... believe me I know the perspective you''re coming from and I respect it. More than you realize.

You cannot discount the fact that the EightStar has highest degree of optical symmetry and is perhaps the finest or one of the finest stones on the market, and you and I both know it. We know that the painting on these unique class of stones is in the highest end of the range 7 degrees azimuth give or take.. from my simulations and constructs..

Regarding craftsmanship Marty I agree 110%. There are few factories on the planet that can cut to the level of precision Allison and her apprentices do. I also like the fact that it is a family owned and operated business just like my own. I SEEK TO SUPPORT OPERATIONS LIKE THIS. At the same time I can''t deny what 6 years of observation testing has shown me and is why I am sympathetic to GIA''s results. I have proposed a resolve to my friends at EightStar which would be beneficial to all parties I believe. If you''d like to know more I prefer to talk about that in private. Regarding the azimuth deviation on EightStar upper halves, the ones I''ve been able to test have ranged from large to very large on Sergey''s scale.

YET..

1) GIA digs it for painting arbitrarily
2) I have YET to see data from ANY scanner that correctly reads the breaks on an EightStar

1) I have not yet found GIA arbitrarily dinging painting. Since the release of the system there have been no changes I am aware of. I saw it stated elsewhere in this thread that GIA was dinging all painting and that they later changed their rules to only ding certain degrees of painting/digging. I''ve been studying the system as in depth as I can for a little over a year now and I''ve never seen them change how they grade painting/digging. There is a systematic approach to their grading on painting and digging that I''ve been able to ascertain. On stones listed on Sergey''s chart with a "moderate" or "practical" degree of painting, these have not been getting dinged in GIA''s system that I''ve found and as I personally observe these stones I do not see any change in optics from stones with neglible/very small/small or moderate degrees of painting.

2) I have 2 scanners that can *see* the breaks. Sarin and Helium. OGI can''t do it yet but I haven''t tested their scanner with the new high definitions camera''s either. I''m seeing if I can talk them into upgrading the cameras in my system for free.

Can you take a Helium scan from an EightStar, which I believe you may have, and present it here anlong with the DMC file.

As long as Richard doesn''t mind me doing this and publishing the fact that its an EightStar.

Sarin scans are currently entirely useless regarding EightStars..

I understand why you say this. Most Sarin''s can''t resolve the breaks in EightStars. Mine can though.

I hope Helium can do the scanning correctly, otherwise how can one quantify with nomenclature what one may not be able to be discerned.. I have been told it can, but I''d like to see few scans..

If its ok with Richard or Dana I personally don''t mind. I have a couple of Helium scans of EightStars I can show ya and I can also show you how our Sarin handles it in various scanning modes.

All the best,

 

[Rhino]

Date: 9/11/2006 11:07:46 AM
Author: strmrdr
is there a maximum? well I guess when there is huge holes left I guess.....
I can live with the list as JohnQ posted in his last post except for maximum

Yea. I prefer huge better than maximum.

 

[Rhino]

Date: 9/11/2006 12:18:22 PM
Author: Serg

re:This one was done the first day he got the helium scanner.
The direct exports contain more data and are more accurate I think.? serg? Garry?

Did Rhino use 400 contours(Accuracy mode) or 800 contours( Hi-Accuracy mode)?
400 contours is not enough for painting 2 clicks. ( Azimuth between main facet and girdle facet is 4 degree only)I have scans done in both modes. I can''t recall off the top of my head which I had forwarded to strm when I did it. I''m seeking permission to post the files. If granted I''ll show ya the 800 contours, otherwise we can take it in email.

Marty,

Important remark : For diamond appearance is important angle deviation only( Slope and azimuth)
Minor facet junction is not important.
If angle between two facet small , even the small angle deviation( error in model) will give very big facet junction errors.
See FS or IS images for this *8star Helium model for example( For classical round diamond such big facet junction errors will give big angle error and change images. But for this example you can not see changing in image)

I''d concur Sergey''s statement here Marty. The extremely minor deviations in the Helium scan have no impact upon the results as all the facets are resolved. If you compare our reflector photography to that of the IS view in the DiamCalc they are about dead on with no notable changes in FS or IS imagery. Here''s an example that demonstrates what Sergey is saying. This photograph is of that EightStar comparing DiamCalc view to detailed photography we took of it. Hope that helps.

 

[Rhino]

oops... here''s the pic.

 

[Rhino]

Date: 9/11/2006 10:55:53 PM
Author: adamasgem

Date: 9/11/2006 10:26:13 PM
Author: canuk-gal
HI:

I acknowledge I am swimming with the big fish here, and hence preface my post by declaring my experience in developing rating scales for research projects, was limited to measuring attitudes.

That said, is it possible to consider using a 7 point linear scale with verbal labels on the end points and the indicators in between labelled with numbers? Diagrams, prn.

Rationale for this suggestion is:

1) 7 points max. based on psychological research by Miller (1956) showing that people have difficulty reliably making more than 7 distinctions (it''s dated, but it''s relevant!). I believe Serg/John mentioned only 7 choices in an earlier post.

2) Research shows that although verbal labels can clarify the meaning of scale points, they may also be distracting. Weisberg et al (1996) surmises it is best to include them, only when necessary. Moreover, if evidence in the literature suggests that people see very little difference between the rating ''Good'' and ''Very Good'', wouldn''t it just be better to use ''the numbers'' and have only the ''extreme'' descriptors/labels on the ends of the scale?

Just an alternative/thought.

cheers--Sharon

As John suggested, a big love note from me

LMAO.

That''s kinda how the former Helium Reports were. The term "classical" was used for terms going up to around the very small range and the "small" variations began getting noted. Also, on the former reports they were not called painting and digging but thinning and gouging. Correct me if I''m wrong on any of that Sergey. So far I like the new details and changes I see. I''ll try to catch up more on this later.

 

[He Scores]

Thanks for clearing things up Sniper, and yes I am now aware of dops that can be fitted with the click wheels in the denominations that Serge pointed out, now that I kinow what we''re talking about.

bill