Optical analysis of mounted diamonds

[Serg]

Date: 12/12/2004 12:55

5 PM
Author: denverappraiser
Serg,

Thank you. That''s just exactly what I''m asking about. You''re amazing. How did you get the wireframe diagram from the picture?

You gave very precise answers but you said before that your accuracy would not be very good. Can you explain? Also, you mentioned that Rhino photos are more appropriate for this kind of work. Why?

Neil

I did it manually. In DC you can load picture and put wireframe double refractive model over this photo.

Firstly You should find Pavilion angle and lover facet. It is easy.

Then table and star facet.

then crown angle . Its very difficult, the type of photo is very important

Then you can find girdle if you know total height.

 

[Serg]

Date: 12/12/2004 1:50:41 PM
Author: denverappraiser
Serg,

If you could take the photograph under controlled circumstances, do you think this technique can produce results that would be useful with the HCA? Sarin data can only be trusted to, at best, 0.2 deg. so you''re only off by a factor of 5 and we have no idea how much of that is attributed to the photography. The more I think about this, the more impressed I am with your work.

Neil

0.2 degree is not big problem for Pavilion angle

For crown such accuracy is very big problem. 0.3-0.4 degree is possible , I think

 

[strmrdr]

Afternoon John.

Im assuming that Neil is looking into including this on appraisals.
Thats why im hard on it.
I wouldnt want it on my diamonds appraisal unless its reasonably accurate.
If there is a way to nail down the crown angles then it might be worth while in some cases.
Also some clarification on using it on non-symetrical diamonds would be useful.

Quoting serg:
"It is possible for symmetrical diamond if great accuracy is not necessary."
If he isnt willing to say its great im sure not. :}

 

[JohnQuixote]

Date: 12/12/2004 3:11:34 PM
Author: strmrdr
Afternoon John.

Im assuming that Neil is looking into including this on appraisals.
Thats why im hard on it.

Afternoon Strm.

I wouldn''t have it any other way from you, mister.

Even if it were reasonably close (not close enough for your paradigms but near what Serg estimates) I imagine it would be an advantage to have this level of info on the parade of pedestrian-quality mounted stones that happen into an appraiser''s shop.

 

[denverappraiser]

Date: 12/12/2004 3:11:34 PM
Author: strmrdr

Im assuming that Neil is looking into including this on appraisals.

Storm

Not necessarily but yes, I'm always interested in improving the quality of my work product. A very common question that comes up here and elsewhere is how to determine cut information on mounted goods. The normal answer is that it's impossible or to use an idealscope, which has no grading scale associated with it. The reasonably accurate measurement tools all require removing the stone from the mounting, which has the risk of damage to both the stone and the mounting and adds expenses that don't really benefit the customer. If there is an optical approach that will provide useful data in a fast, inexpensive and completely non-destructive manner, yes, I'm interested, even if it has a greater margin for error than would otherwise be desirable. There are other limits imposed by the mounting as well but this does not prevent this from being a very useful technique. I love my Sarin machine, but it is usable on less than half of the appraisal assignments. I have no doubt in my mind that this information will be extremely handy. For appriasers that don't have a Sarin machine, or are in a circumstance where it's not possible to use one, this is at least a partial answer to the cut questions that are otherwise impossible to answer.

Neil

 

[Serg]

very big tilt

Lover facet 78
Pav 40.75
Cr 34.5
Table 55.5
star 51.5

8 star diamond I think.

Bye bye

 

[JohnQuixote]

Wow.

Serg you are a master.

I will post exact specs tomorrow, but those sound close, if not dead-on.
So, Serg, you get an A with one exception.

"A Cut Above," not Eightstar.

 

[denverappraiser]

Serg,

Are you using an unmodified version of DC 2.3.1.325 for all of this? Wow

Can you tell us something about the attributes that make a satisfactory picture for this application?

Neil

 

[JohnQuixote]

I do have the IS and H&A images for that diamond on hand.

 

[JohnQuixote]

H&A

 

[Rhino]

Now you can see why Serg thought it was an 8*. There is one primary difference (well... may not be considered primary) but 8*'s star facets are even shorter. You can see this in the measurements plus you can also see the optical effects as well. 8*'s *inner eye* comes to a perfectly pointed star in the middle while the ACA new line doesn't. This is only something that can be observed under magnification through an IS/FS/LS though. Lower girdles & upper girdles are basically identical and I would consider both to be of the shorter star/lower girdle combo. Amazing stuff Serg. For clarification Serg you're listing lower girdle depth as opposed to lower girdle length correct?

 

[Garry H (Cut Nut)]

Neil you need not read the article - it is boring.
It would be in GIA''s library though.

Sergey the viewing distance needs to be set sometimes to match the camera placement and lens type with some known stones.
Options>advanced>viewing distance and then adjust the viewing distance.

You put the picture of the stone into DiamCalc with the background color icon just above the light globe on left tool bar. >set picture.

This is also useful to create nice advertising images

 

[JohnQuixote]

Date: 12/12/2004 6:16:11 PM
Author: Rhino
Now you can see why Serg thought it was an 8*. There is one primary difference (well... may not be considered primary) but 8*'s star facets are even shorter. You can see this in the measurements plus you can also see the optical effects as well. 8*'s *inner eye* comes to a perfectly pointed star in the middle while the ACA new line doesn't. This is only something that can be observed under magnification through an IS/FS/LS though. Lower girdles & upper girdles are basically identical and I would consider both to be of the shorter star/lower girdle combo. Amazing stuff Serg. For clarification Serg you're listing lower girdle depth as opposed to lower girdle length correct?

Rhino - thanks for the obs. You have a lot more experience with 8* than we do so I'm sure you're right. I believe you can see those optical differences with a H&A viewer as well as IS/FS/LS though, right?

Stepping aside

Since you mention it, I hope you know new line's intention was not to emulate 8*. It was produced to demonstrate that limited light leakage could be achieved on our terms, without replicating 8* proportions. 8* is a distinctive and beautiful diamond which is perfect for some peoples' taste. However - you said it right - Brian has never produced facets that short. His design for visual balance has remained true for all A Cut Aboves.

Not really the point of this thread - but if you wish to discuss this more let's start a new thread Sir Rhino (it's a favorite topic of mine).

Steppin' back in

 

[Richard Sherwood]

Neil, I use the DiamCalc on every mounted diamond I appraise over 0.50 carat. Here's the method I use, which requires time but is surprisingly accurate:

(Most measurements are obtained under the microscope using a digital micrometer)

1. Obtain diameter mm by direct measurement
2. Obtain depth mm by profile (sighting) measurement
3. Obtain table mm by direct measurement
4. Obtain girdle mm by direct measurement
5. Obtain crown height by profile (sighting) measurement
6. Don't worry about measuring the pavilion

I input these measurements, and then size the culet on the DiamCalc image to match what I see through the loupe.

The measurements above will usually put you in the neighborhood of actual measurements, but require tweaking for more accuracy. This I do by switching the DiamCalc image to the simulated Idealscope image, and then:

1. Tweaking lower girdles to match what I see through the loupe
2. Tweaking stars to match what I see through the loupe
3. Tweaking the pavilion angle to match what I see through the loupe
4. Tweaking the crown height and girdle thickness, if necessary.
5. The pavilion measurement will come into line as a result of the above work

By this point the Idealscope image is usually looking dead on to the actual diamond (depending on symmetry), and the DiamCalc numbers are far closer than what you could obtain by any other method on a mounted stone. If your Idealscope images are closely matching, you know you've got to be close. I then insert a disclaimer that "on mounted diamonds the proportion numbers are created through an analyis generated by the state-of-the-art Russian DiamCalc software using data input from direct measurement and Idealscope image comparison".

At first it seems to take forever, but after a while it becomes second nature and you get quicker. I feel like the time is worth it, as you are providing information that practically no one else supplies on mounted stones. Not too mention that once you obtain this greater accuracy, it's hard to settle for the older, less accurate method anymore.

I remember Dave Atlas saying, "ANY (accurate) information is better than no information". Most appraisers can't even supply an educated guess on the proportion numbers of a mounted stone, while an appraiser adept with the Idealscope and DiamCalc can supply surprisingly accurate numbers.

Did you notice how Serg nailed the weight and all important pavilion angle, not too mention "smart bombing" the other numbers?

 

[Nicrez]

How funny I find this post AFTER I just used the Liddicott method for my GG 5 stone exam on Friday! If only I had an idealscope to bring to class!!!

Actually, I am glad Neil asked this question, because I was wondering when people where going to use computers for estimation in mounted stones, as the eyeballing method, and the table gauge seem to be so old fashioned that it barely reflects an accurate measurement anymore...

Serg, how does this program account for the girdle, I wonder? Also, how does it calculate the information on a poorly symmtrical stone (such as the monstrosities that are GIA''s student grading stones)?

Does this computer analysis only work on H&A, 8* and equally well cut stones, or does it work with the standard junk seen in old stones? Any usefulness on OECs at all?

 

[Garry H (Cut Nut)]

Date: 12/12/2004 11:48:10 PM
Author: Nicrez
How funny I find this post AFTER I just used the Liddicott method for my GG 5 stone exam on Friday! If only I had an idealscope to bring to class!!!
so they do still teach it Nic - strange since the accuracy requirements are now so much more stringent!

Actually, I am glad Neil asked this question, because I was wondering when people where going to use computers for estimation in mounted stones, as the eyeballing method, and the table gauge seem to be so old fashioned that it barely reflects an accurate measurement anymore...

Serg, how does this program account for the girdle, you must estimate the girdle the old way Nic. It is not apparent from a front view - but if you take a side view photo you cana do the same matching estimation
I wonder? Also, how does it calculate the information on a poorly symmtrical stone (such as the monstrosities that are GIA''s student grading stones)? There is a point when symmetry gest too bad - but if you look back at the blue Tiffany image that I mathced you can see that the stoen was not face on and the symmetry is not so hot - but it can still work

Does this computer analysis only work on H&A, 8* and equally well cut stones, or does it work with the standard junk seen in old stones? Any usefulness on OECs at all? OEC''s or some other fancies are ok too - but princess is difficult because you have the two pavilion angles and variable numbers of "chevrons" or minor facets

 

[valeria101]

Date: 12/12/2004 9:21:24 PM
Author: Richard Sherwood

By this point the Idealscope image is usually looking dead on to the actual diamond (depending on symmetry), and the DiamCalc numbers are far closer than what you could obtain by any other method on a mounted stone.

If your Idealscope images are closely matching, you know you've got to be close.

Neil answers to Strm above: "[...] it's impossible or to use an idealscope, which has no grading scale associated with it."

I am not sure I understand why the "Scope cannot come with a usable scale, once the experiments on this thread were possible. Too debatable ?

 

[Serg]

Re: For clarification Serg you''re listing lower girdle depth as opposed to lower girdle length correct?

In DC we use depth . See:

http://www.octonus.ru/oct/products/helium/parameter05.phtml
Re: "A Cut Above," not Eightstar.
Re: Now you can see why Serg thought it was an 8*. There is one primary difference (well... may not be considered primary) but 8*''s star facets are even shorter.

You can see very bad alignment between model and photo in crown girdle facets. this implyies the azimuth of crown girdle facets is not classical, like in EightStar diamond.. May be "A Cut Above," has same azimuth now, I do not know.

Are you using an unmodified version of DC 2.3.1.325 for all of this?

Yes.

Can you tell us something about the attributes that make a satisfactory picture for this application?

Try and you can find your rules, The main rule : Photo should has more information , specially in main crown facet.
We can not use more crown girdle facets( see above) ./ May be best photo is in the Al Gilbertson light with telecentric lens.

 

[Serg]

Date: 12/12/2004 11:48:10 PM
Author: Nicrez
How funny I find this post AFTER I just used the Liddicott method for my GG 5 stone exam on Friday! If only I had an idealscope to bring to class!!!

Actually, I am glad Neil asked this question, because I was wondering when people where going to use computers for estimation in mounted stones, as the eyeballing method, and the table gauge seem to be so old fashioned that it barely reflects an accurate measurement anymore...

Serg, how does this program account for the girdle, I wonder? Also, how does it calculate the information on a poorly symmtrical stone (such as the monstrosities that are GIA''s student grading stones)?

Does this computer analysis only work on H&A, 8* and equally well cut stones, or does it work with the standard junk seen in old stones? Any usefulness on OECs at all?

If you know real total height and found culet, pavilion angle, crown angle and table by photo then girdle is result.

 

[denverappraiser]

You are all supplying a wealth of information. Thank you.

Obviously this is effective and DC is an enormously useful product. This whole procedure seems to revolve around taking appropriate pictures and accurately attaching the wireframe diagram. This conversion to the diagram seems like something that would automate nicely with some of the photo recognition software but perhaps this is more tricky than I''m expecting. I"ve never worked with any of these products. Aren''t the facet junctions on the stone sufficiently distinctive elements in the photo that, given a tidbit like the basic facet pattern, the program should be able to make a pretty good guess where to locate them.

I''m still having a difficult time conceptualizing a solution to the paralax problem and how serious it is to have a photo from a direction other than exactly perpendicular to the table or how serious these issues are. Practice will probably help.

My problem with the idealscope for this purpose is that there seems to be no very good way to describe the image in a document. Statements like ''excellent'', ''pretty good'' and ''icky'' IS images are not precise enough to be useful for an insurance company to make a replacement (the usual stated purpose of the report). Photos are problematic because IS photographs can vary tremendously depending on the conditions under which the photo was taken and the skills and equipment of the photographer, none of which has anything to do with the merits of the stone. The company has no way to tell the difference between an excellent photographer with a crappy stone and visa-versa.

Neil Beaty
Independent Appraisals in Denver

 

[Serg]

Re:Can you tell us something about the attributes that make a satisfactory picture for this application?

H&A viewer photo is much better than IS/LS/FS photo for such work.

photos from Rhino site.
http://www.goodoldgold.com/2_00ct_e_si1_h&a.htm

 

[JohnQuixote]

Date: 12/13/2004 5:45:30 AM
Author: Serg

You can see very bad alignment between model and photo in crown girdle facets. this implyies the azimuth of crown girdle facets is not classical, like in EightStar diamond.. May be 'A Cut Above,' has same azimuth now, I do not know.

Serg, the azimuth of the girdle facets may not be classical, but it’s our own configuration, not 8*.

You may remember when claims were made that no other stones could be cut with 8*’s light leakage footprint… This is one of the “new line” ACA, cut to retain Brian’s H&A patterning while also showing that such a footprint can be achieved in other configurations.

A question: One of the fundamental problems in photographing the crown-up position is how level you get the stone. It’s time-consuming for loose stones and seems like it would be even harder for mounted, especially in settings with wear and tear.

Is there a way the program can take into account stones that are not 100% level with the camera’s eye?

 

[JohnQuixote]

Date: 12/13/2004 8:50:35 AM
Author: denverappraiser
I'm still having a difficult time conceptualizing a solution to the paralax problem and how serious it is to have a photo from a direction other than exactly perpendicular to the table or how serious these issues are.

Neil - Just saw this. We're on the same page... Do you have any strategies to share?

 

[valeria101]

Date: 12/13/2004 11:37:13 AM
Author: JohnQuixote

Is there a way the program can take into account stones that are not 100% level with the camera’s eye?

If all you need is to fit a wireframe over the picture there may be some way based on rather simple geometry. Once a 2D picture is taken of the tilted diamond, the resulting image should actually represent a projection of the desired level image to a plane of unknown inclination... This projection is an elipse with proportions determined by the inclination of the projection plane: so once you decide how to measure the deformation (this is a problem, since actual symmtery issues get confused with the tilt), you can compensate for the tilt by projecting the image at the respective angle.

Since there are three tilt dirrections to determine (= pitch, yaw and roll) the same trick should be done with several symmetry axis around the diamond... but there is no shortage of that. At least using some automated procedure, one can pick an infinite number of sections. Otherwise, the symmetry of the RBC model already has four such pairs of axis defined for starters.

This simply describes how 3D rotations are generated anyway. Since the wireframe can be rotated in GA, it could be made to match the tilt, right ? Yet more visual tweaking, if acceptable.

If I were to use some photo editing software (Corel), than I may start by choosing a set of key refference points on the immage to calculate the tilt. The visible meet points and the culet can serve for this. Perhaps their determination on the picture can be automated, but this does not sound straightforward at all.

 

[denverappraiser]

Ana,

The problem I'm imagining is that all of those things MAY be problems with the cut. The table is not necessarily centered or parallel to the girdle. The culet may be off-center. The stone may be out of round. I would expect a combination of these sorts of problems to look just like a tilted stone in the photo. Since this is exactly the sort of problem we’re hoping to discover, it seems pretty important. Both Sarin and Ogi use the table plane as the standard from which all other things are measured. This seems like the obvious choice for this approach as well.


John,

No, I don’t have any great ideas. You are a far better photographer than I. The closest I’ve come up with is to have a camera that is pointing straight up from a fixed position below a frame that is, say, 3 feet tall. Set the diamond table down on a piece of glass that is fixed to the frame and take the picture through the glass. This should get reasonably consistent, if artistically mediocre results.


Neil

 

[Garry H (Cut Nut)]

tilt the wire frame to match the tilt of the stone.

Ideal-scope and realistic photo''s can be done quickly and easily with this http://www.ideal-scope.com/manuf_iscapture.asp setup and the autocropping software.

The camera paralax is not a huge problem - it is a minor modification that you can make as you work with stones of known parameters.

But if you use Rich''s method - you do not need the wire frame approach - just match the diamcalc image to what you see in the ideal-scope. This too is very accurate.

 

[PhillipSchmidt]

Edited: wrong post

 

[denverappraiser]

OK Gurus. Here's an IS image taken from a very reproducable environment. The table is reasonably close to perpendicular, the distance and lighting are repeatable. I can do the same with a h&a but I was playing with an IS so I thought I would show the progress. I'm still not wild about the image but it's not too bad. Is this sufficient for the Diamcalc magic?

Garry, care to comment on the IS photo?

Neil

 

[Garry H (Cut Nut)]

Neil can you give me the proportions?
Or photograph a very symmetrical stone whose proportions you have, along with an indication of the thickness at the girdle bones and girdle mains (ie. are they the same -or is one thicker than the other?)

 

[Garry H (Cut Nut)]

But give us the diameter and depth or depth % and I will do this stone also