star and lower girdle cutting precision

[sleeprequired]

I was reading an article on GOG regarding star angles and lower girdle angles... i think i've got that down but the article also goes into the precision and consistency of the cut of all the facets.

so lets say a diamond has star angles of 51-51.6 and lower girdles of say 78.4 - 79.0. What would the real world effect be versus a stone with say 51.0 - 51.2 stars and say 78.8-79.0 ....

my question is what would the lack/increase in precision yield to a keen observer but not an EXPERT as such?

 

[John P]

A small correction: Your numbers don't represent angles, they're percentages of how far from table to girdle (star) or girdle to culet (lower halves) the facets run. This becomes relevant, below.

so lets say a diamond has star angles of 51-51.6 and lower girdles of say 78.4 - 79.0. What would the real world effect be versus a stone with say 51.0-51.2 stars and say 78.8-79.0 .... my question is what would the lack/increase in precision yield to a keen observer but not an EXPERT as such?

On it's face your question is extremely fractional. Forget normal observers or even pros. This is similar to asking what the observable difference would be from FL to IF clarity, or high D to center of D color.

But if you prefer a reply fueled by my own OCD I'd start by naming some prerequisites to your question. First, remember that all of the facets on a diamond work together. Each pavilion main has a lower half on either side. This means the angle of each individual pavilion main has an influence on the resulting angle/percentage of its neighboring lower halves. So variability among the pavilion mains can influence the appearance (or not) of variability in the lower break facets when you're looking at raw numbers. In the crown there is even more variability possible for the stars, with eight mains and sixteen upper break facets. This is one reason we describe the crown and pavilion main facets using angles, while the minor facets get percentages. The "baseline" for each minor can be a bit different depending on the surrounding facets, so it's not apples to apples - strictly speaking - at the most micro-level. And we have not even discussed the possibility of culet off-center, table tilted and/or off-center or girdle variations...

On top of this let's keep scan tolerances in mind. The commonly accepted error envelope is 0.2 degrees angular and 0.5% linear for a calibrated scanner, with wiggle room for facets adjacent to the table. Things have improved notably amongst 3D scanners in recent years but they are still not perfect. So when considering numbers at an extremely micro-level remember that cross-scanner and cross-platform comparisons, especially, must be taken with a grain of salt.

If your question is about 3D precision & consistency, a professional-caliber hearts & arrows viewer can demonstrate more about optical 3D precision than a scan result, since it shows more than raw numbers which may not be conclusive when taken singly. Note that I didn't say a H&A photo does this. Why? Because resolution, focal depth, lens-plane alignment relative to table and other macro-challenges make general evaluation possible but subject to limitations. I wouldn't make a decisive H&A call based on a photo any more than I would make a color or clarity call based on a photo. I can tell you a lot from a good photo, but at the highest micro levels an in-person evaluation using a high quality reflector with a pro lens is necessary to discern the details.

Notice I am talking about optical 3D precision results in the above paragraph and not wire-frame 3D precision. The two can be related but are not necessarily always hand-in-hand. The latest innovation in assessing Wire-frame 3D precision is discussed here.

https://www.pricescope.com/community/threads/ags-lab-releases-vpa-software.175050/

 

[Rhino]

I was reading an article on GOG regarding star angles and lower girdle angles... i think i've got that down but the article also goes into the precision and consistency of the cut of all the facets.

so lets say a diamond has star angles of 51-51.6 and lower girdles of say 78.4 - 79.0. What would the real world effect be versus a stone with say 51.0 - 51.2 stars and say 78.8-79.0 ....

my question is what would the lack/increase in precision yield to a keen observer but not an EXPERT as such?

I'll cut to the chase. Whether it was a keen observer OR an expert ... with differences so minute the real world effect would be NADA.

 

[Karl_K]

The scanners are not accurate enough to play this game.
You could get both those results from the same stone on the same scanner.
The lgf% is one of the harder measurements for a scanner to get right.

 

[yssie]

The scanners are not accurate enough to play this game.
You could get both those results from the same stone on the same scanner.
The lgf% is one of the harder measurements for a scanner to get right.

If it was a repeatable bias error you could calibrate and correct

Mr. John P. - surely an error at the limit of a stated a 0.2deg uncertainty (by whatever confidence/std dev) would be a sufficiently large deviation to affect not only results but even result trends of what is undoubtedly a nonlinear bowl of noodles, no? What comes out can only ever be as accurate as what goes in... still useful info as long as the user *fully understands* the constraints and initial assumptions!

 

[Karl_K]

If it was a repeatable bias error you could calibrate and correct

I wish it was that easy but it is not.
Something to think about no facet cut by man is ever going to be perfectly flat and square and have the same angle across its face not to mention perfectly aligned in 3D space with no twists or yaw.
Which point is the angle of the facet to report? The highest? the lowest? average?
So you have an imperfect tool measuring an imperfect item.

 

[John P]

Mr. John P. - surely an error at the limit of a stated a 0.2deg uncertainty (by whatever confidence/std dev) would be a sufficiently large deviation to affect not only results but even result trends of what is undoubtedly a nonlinear bowl of noodles, no? What comes out can only ever be as accurate as what goes in... still useful info as long as the user *fully understands* the constraints and initial assumptions!

Exactly. GIGO. Which is why I believe in-person visual examination is necessary for nitpicking optical 3D precision.

 

[yssie]

Karl - average, round, and average again around the stone. Obviously.

But, that's the point of UQ - to acknowledge that there is error that you can't correct for, and to try to quantify its effects or at least qualify the trends of its effects in some meaningful way. And I am persnickety about UQ. I'm not the least bit offended that the machine is less accurate or unreliably inaccurate on certain facets or structures; I'm unimpressed that this is all the Sarin website says, no further info or even definition of what population proportion that +/- represents (for the DiaMension):
Linear: ± 20 microns (± 0.02 mm)
Angular: ± 0.2 degrees

I think the new AGS software looks really nifty - certainly gives really nifty printouts - I'm staying far away from the details, I'll leave that to braver and more knowledgeable folk!


John - thanks for responding, and of course you are correct!

ETA sleep sorry to have dragged your thread OT! Letting it go now...

 

[sleeprequired]

Ok so as a purchaser if a sarin reads 34.5 - 35.5 it's plus or minus say 0.2 so it could be a variation of 0.6 at best or 1.4 at worst.

Still if you get the chance better to pick a stone where the sarin says 34.5-34.7 surely?

Having said that the difference to the look and performance of the stone would still be insignifficant even to the trained eye? No effect on brilliance/scint/fire etc.

Best really to ask the vendor "which one is better to the eye?" and trust them than worry about trying to work it out?

Assume both stones still show good IS and ASET and are ideal everything accord the board.

 

[yssie]

Ok so as a purchaser if a sarin reads 34.5 - 35.5 it's plus or minus say 0.2 so it could be a variation of 0.6 at best or 1.4 at worst.

Still if you get the chance better to pick a stone where the sarin says 34.5-34.7 surely?

Having said that the difference to the look and performance of the stone would still be insignifficant even to the trained eye? No effect on brilliance/scint/fire etc.

Best really to ask the vendor "which one is better to the eye?" and trust them than worry about trying to work it out?

Assume both stones still show good IS and ASET and are ideal everything accord the board.

That's the point.
A) If the scanner that they use has stated error of +/- 0.2deg, if a given measurement is 34deg does that mean the true value is within an absolute lower bound of 33.8 and an absolute max of 34.2? That the range of 33.8-34.2 represents one standard dev differential? Confidence interval? Someone in the trade may have more info on this.
B) If the scanner error is such that the first time the stone is scanned it shows 34.5-34.7 and the second time the same stone is scanned by the same scanner it shows 34.3-34.5 or 34.3-34.7 it makes no sense at all to quibble over those sorts of differences.


It is absolutely better to just ask your trusted vendor about which looks better, if either, and have them articulate why!

 

[sleeprequired]

On it's face your question is extremely fractional. Forget normal observers or even pros. This is similar to asking what the observable difference would be from FL to IF clarity, or high D to center of D color.

Thanks John, I actually took a look at the D vs G colour comparison video on GOG where the two stones were effectively identically cut.

http://www.youtube.com/watch?v=1DJaELN6xgY&ad=8132676332&kw=diamond+information&lr=1&feature=pyv

If you're saying that the differences in precision in the example i gave in the opening post are like high D to mid D colour, then that really helps me put these numbers in perspective because even side on there's not a massive difference from D to G. Comparing that to your analogy, there would have to be a massive difference in cut precision to cause a problem, not simply a 1.4% difference from minimum to maximum including the +- 0.2 accuracy.

the angle of each individual pavilion main has an influence on the resulting angle/percentage of its neighboring lower halves. So variability among the pavilion mains can influence the appearance (or not) of variability in the lower break facets when you're looking at raw numbers

So the pavilion angle variance will cause a variance in the lower girdles? am i reading that right? and the crown angle variation will affect the stars?

the possibility of culet off-center, table tilted and/or off-center or girdle variations..

is this something that a ags tripple ideal would weed out?

Thanks for all your help too

 

[John P]

If you're saying that the differences in precision in the example i gave in the opening post are like high D to mid D colour, then that really helps me put these numbers in perspective because even side on there's not a massive difference from D to G. Comparing that to your analogy, there would have to be a massive difference in cut precision to cause a problem, not simply a 1.4% difference from minimum to maximum including the +- 0.2 accuracy.

Yes.

I have more comments, but yes. Whenever it gets really micro-tech just remember the above.

Yes.

the angle of each individual pavilion main has an influence on the resulting angle/percentage of its neighboring lower halves. So variability among the pavilion mains can influence the appearance (or not) of variability in the lower break facets when you're looking at raw numbers

So the pavilion angle variance will cause a variance in the lower girdles? am i reading that right? and the crown angle variation will affect the stars?

It doesn't create variance, it creates different baselines. Ergo the "raw" number for any given minor facet (even if the scanner was 100% accurate) can't be strictly compared to the next. Example: If pavilion main #1 has a 40.6 degree angle and pavilion main #5 has a 40.9 degree angle the adjacent lower halves could have been polished identically, but the lower halves touching #1 will read as having a different "raw" angle than those for main #5.

In classroom terms: southbound trains A and B are identical, traveling at identical speeds. They depart for Houston. If they leave the same station traveling the same speed they will arrive at identical times. But if A leaves from Dallas and B leaves from Ft Worth the "raw" result will be that train A is faster - when actually it's identical, considered in full-context.

the possibility of culet off-center, table tilted and/or off-center or girdle variations..

is this something that a ags tripple ideal would weed out?

Thanks for all your help too

You're welcome. And yes, but even the highest cut grades have a permitted tolerance. It would be impractical for them not-to. Some degree of off-center T/C and give and take in polishing will occur with any diamond. It is the hardest known natural substance but even diamond has softer and harder places as well as graining and natural characteristics which all influence the crafting. In that sense we are so off the chart with the micro-considerations in this thread that I wager you'd have a hard time finding jewelry pros IRL who can (or will) engage in discussion at this level. Welcome to Pricescope.

In that spirit, I'm going to point again to your first text above, highlighted in red. That overrides everything else in this thread.

 

[John P]

A) If the scanner that they use has stated error of +/- 0.2deg, if a given measurement is 34deg does that mean the true value is within an absolute lower bound of 33.8 and an absolute max of 34.2? That the range of 33.8-34.2 represents one standard dev differential? Confidence interval? Someone in the trade may have more info on this.

If the scanner is well-maintained and calibrated it's the lesser of the two: Ocular error is likely be in the same physical hardware direction, although the results can't be "corrected" since facet orientation through rotation will seem to spread that error around.

B) If the scanner error is such that the first time the stone is scanned it shows 34.5-34.7 and the second time the same stone is scanned by the same scanner it shows 34.3-34.5 or 34.3-34.7 it makes no sense at all to quibble over those sorts of differences.

Such chaos would be a calibration or software issue. And while it's simple to overcome this by acquiring updates and calibrating the hardware at the beginning of every period of use I am shocked at the number of entities I encounter who have not calibrated the machine since getting it...and you'd have to get in the wayback machine to find the version of the software being used.

But with all that said, it's still not perfect. And the same scanner can indeed produce small but differing results for the same diamond if run twice.