IGI, GiA, AGL-gem labs

[Serg]

So Sergey, based on say a rectangular clean LGD block of say 14x10x7mm you could propose emerald cut proportions that a skilled manufacturer can produce an exceptional 5ct stone?
(There is no need to fight for yeild with the LGD material)

Short Answer: Yes.
Detailed answer :
We are working on creating an Optical Performance evaluator for Emerald cuts. This evaluator not only has a Grid part like Facetware but also includes preforms (around 50 3D models) with a high Brilliancy score (approximately 0.9 for emeralds).

The problem is that changing the angle of certain facets by 0.5 degrees can create DeadZones that significantly reduce Brilliancy. Even the most experienced cutter, using the best technology available today, cannot achieve an accuracy and repeatability better than 0.5 degrees on emerald cuts (an error of 1 degree is quite common).

Therefore, GRID optimization is not very effective for emeralds, and we will be converting the GRID evaluator into a Fixed Cuts system. Developing one evaluator takes about one year.

 

[Karl_K]

Since this has side tracked into production.
If I was producing cvd LGD rounds I would pick 1 proportion set and automate the entire process as much as possible from growth to final finish to that 1 proportion set and 3 or 4 sizes.
Pick a combo all the labs agree on and floor it.
Multiple proportion sets in the process makes little sense with cvd LGD in my opinion.

 

[smitcompton]

Hi,

Since this has side tracked into production.
If I was producing cvd LGD rounds I would pick 1 proportion set and automate the entire process as much as possible from growth to final finish to that 1 proportion set and 3 or 4 sizes.
Pick a combo all the labs agree on and floor it.
Multiple proportion sets in the process makes little sense with cvd LGD in my opinion.

Hi,

I so see the reasonableness of this market approach. Frankly, it is one of my problems with ACA diamonds. They look as if they came out of an automated factory line. They all look the same, To my uninformed view I sort of say to myself, these must be laser cut. But LGD should be automated .

Garry-- I was thinking of EGD when I proposed DeBeers as a sponsor of a contest. But I see, it would more practical to have new cuts developed with the less expensive material.

Annette.

 

[Garry H (Cut Nut)]

Garry-- I was thinking of EGD when I proposed DeBeers as a sponsor of a contest. But I see, it would more practical to have new cuts developed with the less expensive material.

Annette.

Yes Annette,
More than a decade ago, with Sergey and others, we came up with a plan to give DiamCalc like cut designing ray tracing software to entrants - promoted to design school students and whoever.
Those who came up with the best designs would be offered to various firms to produce the diamonds, with some licenced payments and exclusivity.
And finally the finished diamonds could be assessed and judged by various judges: consumer fav, and any other authority or association. there may be multiple award winners and many new products. Designers and manufacturer's can all win.
Natural diamond winning styles may well be deeper portioned so as to work well with natural rough, etc etc
@Serg - is this a fair summary of the old plan?

 

[Ibrakeforpossums]

Since this has side tracked into production.
If I was producing cvd LGD rounds I would pick 1 proportion set and automate the entire process as much as possible from growth to final finish to that 1 proportion set and 3 or 4 sizes.
Pick a combo all the labs agree on and floor it.
Multiple proportion sets in the process makes little sense with cvd LGD in my opinion.

Lightbox is grading in batches now. This would do them one better.

 

[V_sh]

Therefore, the parameter-based approach and 3D models are quite effective for cutters, unlike in the case of laboratory tasks. For cutters, a false negative error does not pose a risk. False positive errors are practically non-existent if the proportions are correctly verified. However, what works well for cutters is not very applicable for laboratory tasks.

Short Answer: Yes.
Detailed answer :
We are working on creating an Optical Performance evaluator for Emerald cuts. This evaluator not only has a Grid part like Facetware but also includes preforms (around 50 3D models) with a high Brilliancy score (approximately 0.9 for emeralds).

Jasper Paulsen is a mathematician interested in diamonds. Jasper improved Tolkowsky’s “ideal” round proportions at www.folds.net.
It could take a century or more to build and validate a database to establish five cut quality grade boundaries. Building a system to enable diamond cutters to plan the most efficient model for individual rough diamonds is then the next complex step.

Dear @Serg and @Garry H (Cut Nut)

The following scores are related to two RBCs and THE ONLY difference between these two stones is just a 0.10° difference in the Main crown facet angle.

@Garry H (Cut Nut) why do you think the calculation that a human (not a computer) has done can solve the cutting/designing problem? and can we understand emerald cut when we haven't understood RBC yet? (or maybe you have developed an algorithm for this task?)

@Serg Will your Optical Performance evaluator for Emerald cut be enough accurate when we haven't understood RBC completely? will it be practical when "an error of 1 degree is quite common"? (or if Octonus knows everything about RBC, please kindly interpret the above scores)


@Serg and @Garry H (Cut Nut) If you can explain the above scores describing two stones that are merely different I'll support your Fancy Cut mindsets.

The computer 3D models are obviously showing that we should increase the accuracy of cutting instruments AND to increase our knowledge about the well-known RBC at first.

Your approaches remind me of the Toyota approach when they entered the F1 world.

Our foundations for understanding diamond performances are not strong enough to enter the Fancy Cut world. (If I'm wrong please generously enlighten me, gentlemen)


@Karl_K your opinion will be interesting, please

 

[Serg]

Natural diamond winning styles may well be deeper portioned so as to work well with natural rough, etc etc
@Serg - is this a fair summary of the old plan?

About ten years ago, we decided to develop a Cushion cut that would have high Fire, Brilliancy, and no leakage. At that time, there were three types of Cushion cuts on the market:

1. Crushed Ice, which are very bright when observed under Trade conditions (bright white background), but dull in closed jewelry settings due to high average leakage caused by chaotic long light paths (the light exits through the crown and pavilion with equal probability). These cuts also have almost no Fire.
2. Cuts with a square girdle shape and brilliance similar to the classic round cut. In my opinion, these cuts lose all the charm inherent to Cushion designs and are essentially disfigured round cuts, resulting in a very high variation in girdle thickness.
3. Dozens of hybrid modifications (Chimeras) that use the crown of the classic round cut and modifications of the pavilion idea for Princess or Radiant cuts. In recent years, these modifications have been continually improved, and now some of them have fairly good Optical Performance, such as Ankit nice example. Interestingly, the same design can produce both large virtual facets (short light paths) and Crushed Ice, depending on the crown and pavilion angle ratios. Diamonds look like different cuts, although they have identical patterns. This is a very interesting phenomenon, and for the Optical Performance courses, we are cutting three samples to show how you can get a Crushed Ice Cushion from a non-Crushed Ice Cushion by changing the proportions.

Among this family of cuts, there are some very interesting and worthy representatives, but I wanted to create a cut with a uniform pattern of virtual facets, where the facets are of optimal size to create Life and Fire, evenly distributed across the cut. In contrast, even in the best samples, there were zones of large and very small virtual facets, disrupting the harmony of the pattern in terms of Fire, Brightness, and dynamic characteristics. A Chimera remains a Chimera.

I believe we managed to create something new and interesting with a uniform distribution of virtual facets and high Optical Performance. This design (Cushion 4) is very similar to the classic RBC in terms of Optical Performance but has a completely different virtual facet pattern, creating an interesting 3D ornament.


However, the spread of this cut is highly negative. If you saw this cut, you would think it was created to achieve a high yield. Yet, for most rough materials, the yield of this cut is worse than RBC, and finding rough where the yield is the same or slightly higher than a round cut is quite challenging. Even using rough where the yield is 5% higher, the diamond looks much smaller than the round diamond cut from the same stone. Despite the fact that many people liked this cut, the large negative spread killed its commercial potential. No one wants to cut a design that simultaneously has a worse yield than RBC and a -20% spread. We then spent several years trying to improve the spread while maintaining Optical Performance, but we were unsuccessful. We had to sacrifice some Optical Performance to improve the spread (make the cut less deep).

I shared this long story to show how erroneous your assumptions are that cutters benefit from cutting deep diamonds. If you create designs to maximize Optical Performance, you can end up with a cut with a large negative spread, which some experts might mistakenly consider designed for high yield, presenting arguments that seem convincing to consumers but have no basis in reality or practice.

Regarding Cutgame, it is much more complicated than just distributing ray tracing software. The key point here is IP protection. Until the problem of commercial design protection is solved, it will not be profitable to develop and promote new designs. Those who will win are not those who invest in development, but those who copy the best designs without bearing any R&D costs.

One of the most practical ways to protect design and investments in manufacturers' brands is to create a cutters' association that will control design rights and sign a collective agreement with leading laboratories to prohibit the certification of diamonds from companies that do not own the design rights.

 

[Serg]

Dear @Serg and @Garry H (Cut Nut)

The following scores are related to two RBCs and THE ONLY difference between these two stones is just a 0.10° difference in the Main crown facet angle.

0.1 degree in Main pavilion facets cannot cause such a huge difference in the Fire score of an RBC. There must be another reason. First, we need to check the differences in clarity and color, as well as whether both diamonds were properly cleaned before measurements. Please provide links to these samples.

 

[V_sh]

If you create designs to maximize Optical Performance, you can end up with a cut with a large negative spread, which some experts might mistakenly consider designed for high yield, presenting arguments that seem convincing to consumers but have no basis in reality or practice.

Thank you so much for sharing your highly knowledgeable opinion; as I always expect, a worthy story is a long one.

Sharing your precious experience helped me a lot appreciate.

The key point here is IP protection. Until the problem of commercial design protection is solved, it will not be profitable to develop and promote new designs. Those who will win are not those who invest in development, but those who copy the best designs without bearing any R&D costs.

@Garry H (Cut Nut), I completely agree with Sergey, at this moment I'm working on further understanding of "Light in a Diamond" Some day I'll start to develop a cut that I really like, IF a day I achieve a nice cut I could surely say that I put more than 5 years on that.

Why should I participate in a cutting/designing game before contracts/patents??

0.1 degree in Main pavilion facets cannot cause such a huge difference in the Fire score of an RBC. There must be another reason. First, we need to check the differences in clarity and color, as well as whether both diamonds were properly cleaned before measurements. Please provide links to these samples.

Sharing your experience-based opinion means a lot to me thank you so much.

0.1 degree change in Main Crown produced the above results, which I can't mathematically explain why.

Clearly, you are pointing to the "basic features of the scientific method" that I'm aware of.

I'll share my findings as soon as they are completed.

 

[Serg]

0.1 degree change in Main Crown produced the above results, which I can't mathematically explain why.

Clearly, you are pointing to the "basic features of the scientific method" that I'm aware of.

I'll share my findings as soon as they are completed.

Why are you refusing to publish the links to this pair of diamonds so that anyone can start studying this example and help find the reason why two diamonds with only a 0.1-degree difference in crown angle have such different Optical Performance scores?

 

[V_sh]

Why are you refusing to publish the links to this pair of diamonds so that anyone can start studying this example and help find the reason why two diamonds with only a 0.1-degree difference in crown angle have such different Optical Performance scores?

The answer is quite simple.

No wise person will share something that he's interested in, with people who are not interested.
I do respect my interests and I do respect a person who doesn't like my interests and doesn't want to waste his time with my things, so I won't share them for the respect of all.

the only interested persons here are you and me, so I'll email you links of the reports.

I also have a question, please:
Is there any (ANY) diamond cut that you know about, which its fire will reduce by 48% ; because of 0.10°change in any angle?

Thank you so much for your time and consideration, Sergey.

 

[Serg]

The answer is quite simple.

No wise person will share something that he's interested in, with people who are not interested.
I do respect my interests and I do respect a person who doesn't like my interests and doesn't want to waste his time with my things, so I won't share them for the respect of all.

the only interested persons here are you and me, so I'll email you links of the reports.

I also have a question, please:
Is there any (ANY) diamond cut that you know about, which its fire will reduce by 48% ; because of 0.10°change in any angle?

Thank you so much for your time and consideration, Sergey.

I support the acquisition and dissemination of knowledge, and if I have the opportunity, I provide the data on which my conclusions are based. I expect the same from others. If someone makes claims but refuses to provide information to verify the basis of these claims, the discussion loses its meaning for me.

On this forum, there are participants with different opinions, experiences, and knowledge. What you can gain from interacting on this forum depends on how respectfully you treat other people's viewpoints, even if you don't agree with them. Perhaps you didn't notice, but you have essentially put me in a position where I must choose between siding with you or with the other forum participants. I choose the forum. I respect courage, not belligerence.

With great regret, I must stop responding to your questions, both public and private.

 

[Garry H (Cut Nut)]

The answer is quite simple.

No wise person will share something that he's interested in, with people who are not interested.
I do respect my interests and I do respect a person who doesn't like my interests and doesn't want to waste his time with my things, so I won't share them for the respect of all.

the only interested persons here are you and me, so I'll email you links of the reports.

I also have a question, please:
Is there any (ANY) diamond cut that you know about, which its fire will reduce by 48% ; because of 0.10°change in any angle?

Thank you so much for your time and consideration, Sergey.

Not true. Many are following V

 

[Karl_K]

@V_sh the free sharing of information is something that has always been a huge part of this site.
One thing I have learned is there are far more people following along than you think interested and learning from it.

One thing to keep in mind is you can apply a zillion different computer tests to a design but you never know for sure what you have until its cut and people look at it.
Hit on something awesome and then if it has weird looking numbers on paper good luck trying to sell it.
There is so much involved.

 

[V_sh]

If someone makes claims but refuses to provide information to verify the basis of these claims, the discussion loses its meaning for me.

Dear Sergey,

Thank you so much for your time and for enlightening me about what you expect and how you think.

I exactly agree with you and I was trying not to discuss this in the forum as I thought it seemed silly for friends that I act and question weirdly; Which our friends proved to me that I was wrong with their generous replies.

For sure I wasn't refusing to share data (sharing is why I'm here for) I just felt that I would better discuss this with you by email rather than in the forum.


and by saying

Clearly, you are pointing to the "basic features of the scientific method" that I'm aware of.

I'll share my findings as soon as they are completed.

the only interested persons here are you and me, so I'll email you links of the reports.

I was trying to listen to your warning about the accuracy of the comparison between diamonds and just wanted to share my complete results with you after doing the experiment again.

I thought it was obvious that I'll share the data soon, why are you mad at me for not sharing? (I'm REALLY confused)


anyway, you may kindly consider that we are from different generations with different cultures and different mother tongues and just typing for each other in a third language without seeing each other’s expressions; this can easily go wrong with misunderstandings, am I right?

so you can have my word that I was not insulting you or other friends; my responses were because I thought that no one was caring, I think you know this forum and how it works better than me because I really couldn't even expect new replies in this forum. (I learned from what @Karl_K and @Garry H (Cut Nut) is saying)

In my culture, I have to highly respect you and your decisions because you are elder and more experienced than me; so I will have to accept your decisions whether I like it or not. (you decide, I'll obey)

You may generously accept my apologies if I bothered you unintentionally.

your replies, and also the replies of other friends are so precious to me and I'm just studying what all of you say, as hard as I can whether I agree or not.

it will be my great honor to have your support if you accept to.

Sincerely

 

[V_sh]

Not true. Many are following V

@V_sh the free sharing of information is something that has always been a huge part of this site.
One thing I have learned is there are far more people following along than you think interested and learning from it.

One thing to keep in mind is you can apply a zillion different computer tests to a design but you never know for sure what you have until its cut and people look at it.
Hit on something awesome and then if it has weird looking numbers on paper good luck trying to sell it.
There is so much involved.

Thank you so much Garry and Karl,

you may not believe me but I really didn't expect anyone following this thing.

Thank you so much for informing me and your valuable replies.

how can I know if anyone is following a threat? Please generously teach me I really don't know

Hit on something awesome and then if it has weird looking numbers on paper good luck trying to sell it.
There is so much involved.

Thanks Karl, It may be interesting for you to know this is not my business and I'm certainly not looking to find something to sell.

I'm a science researcher and have my own career with no relevance to diamonds and this is just my scientific hobby as I enjoy and I'm good at Mathematics, Physics, and problem-solving.

so the diamond cutting is just like having an Aquarium for me (but I'm serious in my hubbies) I just enjoy learning from all of you and exploring things.
for sure I won't hesitate to make money from my hobby but that's not my purpose.

and don't look at the "Trade" mark under my name; I just ticked it "yes" because I do some colored gem selling for fun, I didn't know this forum may be this serious when I signed up a few years ago. LOL

 

[V_sh]

So let's get down to the main topic. (please consider 3D virtual method I used)

I'll inform you what I did based on each step that I performed.

Based on what I saw It seems to me that for each light performance quality (Fire, Brilliance, etc) and for each cut there is a shifting point related to the facet(s) that cause that light property, some small changes like 0.05° may cause a big amount (10%) of decrease or increase in fire (I saw this in LG changing for RBC)

please check the following reports of the first step and let's talk about it.

do you know why small changes in angle make this difference in a single 8/8 cut?

EX Round, 1.00ct, D, IF, EX, EX, NON, +1.4%

EX Round, 1.00ct, D, IF, EX, EX, NON, +1.4%

cutwise.com

EX Round, 1.00ct, D, IF, EX, EX, NON, +1.4%

EX Round, 1.00ct, D, IF, EX, EX, NON, +1.4%

cutwise.com

@Serg I thought it may be an error from the system but after checking some other ranges of different groups of facets of RBC I faced some other big changes (0.05° in LG I mean), you may kindly let us know your opinion based on your technical knowledge from your 3D virtual method

@Garry H (Cut Nut) do you see such shiftings in other cuts? or maybe a valuable experience you have from HCA developing, please

@Karl_K I know this is just computer thing, but can we have this as physicall stones? the accurecy of cutting I mean; do have any valuable experience in comparing two closely related physicall stones,please?

The experiment with single cuts showed me how wothless it can be to compare cuts based on MP, MC, and Table only, do you agree?

I'll wait for your precious replies, appreciate

 

[Karl_K]

@V_sh
Rather than feed you the answers im going to help you find them.
Go to different RB stones on cutwise go down to "html report" and view the deviation columns then and pay attention to the min/max angle ranges for each facet set.
Start with these:

EX Round, 2.30ct, E, VVS1, EX, EX, FNT, -5.7%

EX Round, 2.30ct, E, VVS1, EX, EX, FNT, -5.7%

cutwise.com

EX Round, 2.19ct, D, IF, EX, EX, FNT, -3.5%

EX Round, 2.19ct, D, IF, EX, EX, FNT, -3.5%

cutwise.com

EX Round, 5.00ct, H, VVS1, EX, EX, MED, -4.8%

EX Round, 5.00ct, H, VVS1, EX, EX, MED, -4.8%

cutwise.com

View several more also.

Edited to add another and say that not all of them have the html report available.

 

[0515vision]

Gentlemn, I only understand about 0.01% of what’s going on, but I love this thread and thanks for all your contributions!

 

[V_sh]

@V_sh
Rather than feed you the answers im going to help you find them.
Go to different RB stones on cutwise go down to "html report" and view the deviation columns then and pay attention to the min/max angle ranges for each facet set.
Start with these:

EX Round, 2.30ct, E, VVS1, EX, EX, FNT, -5.7%

EX Round, 2.30ct, E, VVS1, EX, EX, FNT, -5.7%

cutwise.com

EX Round, 2.19ct, D, IF, EX, EX, FNT, -3.5%

EX Round, 2.19ct, D, IF, EX, EX, FNT, -3.5%

cutwise.com

EX Round, 5.00ct, H, VVS1, EX, EX, MED, -4.8%

EX Round, 5.00ct, H, VVS1, EX, EX, MED, -4.8%

cutwise.com

View several more also.

Edited to add another and say that not all of them have the html report available.

Thanks Karl,
yeah I usually do so, and some times check the angles of all facets if the varience is great. (non symmetrical ones)

How can this help us?

 

[Karl_K]

Thanks Karl,
yeah I usually do so, and some times check the angles of all facets if the varience is great. (non symmetrical ones)

How can this help us?

When a deviation of .2 on pavilion mains is world class and anything under .3 is tight.
For the crown angle, lower girdle,upper girdle facets its higher than that.
0.05 degree becomes a slight rounding error and .1 degree total deviation a dream.

There are some step cuts where .05 degrees becomes very important on some facets but with an MRB its a rounding error.

Then when you consider GIAs gross rounding .1 is a rounding error too.

 

[V_sh]

When a deviation of .2 on pavilion mains is world class and anything under .3 is tight.
For the crown angle, lower girdle,upper girdle facets its higher than that.
0.05 degree becomes a slight rounding error and .1 degree total deviation a dream.

There are some step cuts where .05 degrees becomes very important on some facets but with an MRB its a rounding error.

Then when you consider GIAs gross rounding .1 is a rounding error too.

Oh yes, for sure I agree

But my opinion is that the cutting instruments must become more accurate and also designs because it makes a good difference if we could be more accurate in cutting

If you remember the topic that made me grab your attention to the small changes in angles, was the amount of possible errors in cutting diamonds.

cannot achieve an accuracy and repeatability better than 0.5 degrees on emerald cuts (an error of 1 degree is quite common).

@Serg told us about errors in Emerald cuts, and I was discussing that even in RBC 0.10 degree matters and we don't understand it yet (at least I think so, am I wrong?), so we can not really talk about the accurate performance in fancy cuts when the error is 0.50 - 1 degree

RBC project has come so far in the last century, but my opinion is that it's not completed yet, and before completing the RBC project we can not consider a long-lasting assumption in fancy cuts. Do you agree?

@Serg @Garry H (Cut Nut) Then the main question that I asked was, why should we pay attention to fancy cuts and put the old classic RBC away when new 3D models are showing us even 0.10 degree can make a big difference but we've never assessed such a thing?

Just because we can not cut diamonds to 0.01° accuracy it doesn't mean we should not pay attention to highly precision cuts (my opinion)

Thanks for your valuable time and explanation, that means a lot

 

[Karl_K]

Oh yes, for sure I agree

But my opinion is that the cutting instruments must become more accurate and also designs because it makes a good difference if we could be more accurate in cutting

If you remember the topic that made me grab your attention to the small changes in angles, was the amount of possible errors in cutting diamonds.

@Serg told us about errors in Emerald cuts, and I was discussing that even in RBC 0.10 degree matters and we don't understand it yet (at least I think so, am I wrong?), so we can not really talk about the accurate performance in fancy cuts when the error is 0.50 - 1 degree

RBC project has come so far in the last century, but my opinion is that it's not completed yet, and before completing the RBC project we can not consider a long-lasting assumption in fancy cuts. Do you agree?

@Serg @Garry H (Cut Nut) Then the main question that I asked was, why should we pay attention to fancy cuts and put the old classic RBC away when new 3D models are showing us even 0.10 degree can make a big difference but we've never assessed such a thing?

Just because we can not cut diamonds to 0.01° accuracy it doesn't mean we should not pay attention to highly precision cuts (my opinion)

Thanks for your valuable time and explanation, that means a lot

The links you shared are not MRB cuts.
They are single cuts.
I also very strongly suspect there is a error with the results.

I posted MRB links to show what kinds of angle deviations are normal in the real world.
Lets pay attention to them when discussing MRBs.

 

[V_sh]

The links you shared are not MRB cuts.
They are single cuts.
I also very strongly suspect there is a error with the results.

I posted MRB links to show what kinds of angle deviations are normal in the real world.
Lets pay attention to them when discussing MRBs.

Yeah single cuts,

Surely you are right, I've been trying to do the experiment again when Sergey said there can be a problem. (although I checked couple of things and ranges before)

Same thing I've seen, when adding new sets of facets
(for example if LG added to the single cut or RBC without UG only)

Maybe we discuss it again if Sergey and Garry help.

 

[Serg]

Cutwise

cutwise.com

At first glance, this pair of cuts demonstrates the instability of the Fire metric on large, low-brightness colored flashes. The problem is determining what should be considered a Fire flash and what should not. Where should the brightness threshold of a flash be?

When assessing the contribution of a flash, we do not use discrete (binary) thresholds; instead, we use weighting coefficients that correlate with brightness and saturation. But before assessing the contribution of a flash, it first needs to be identified in the diamond image, and this is where the problem of defining what constitutes a colored flash and its boundary arises (some metrics do not have this problem as they simply count colored pixels, but in that case, they cannot account for the linear size of the flash).

In this cut, there are very dark blue Fire flashes with a soft brightness gradient on the pavilion facets under the table. It is very difficult even for the eye to determine where the colored flash ends. Usually, such dark colored facets appear in zones of partial leakage, where, for example, red and green rays exit through the pavilion, while the blue rays, due to their higher refractive index, partially remain in the diamond. Most Fire flashes are created by the eye cutting off part of the spectrum from a conical beam exiting the diamond, typically resulting in bright Fire flashes.

A more detailed analysis may take several weeks. However, it is already clear that a 0.1-degree difference does not create a significant difference in Fire for these cuts. The 0.1-degree difference highlights the rounding error in determining what is considered a Fire flash.

Whatever the criterion or threshold for what is considered a Fire flash, it is always possible to find a cut that creates flashes with this borderline threshold. Consequently, the slightest change in the cut will move such flashes from one category to another

 

[V_sh]

Cutwise

cutwise.com

At first glance, this pair of cuts demonstrates the instability of the Fire metric on large, low-brightness colored flashes. The problem is determining what should be considered a Fire flash and what should not. Where should the brightness threshold of a flash be?

When assessing the contribution of a flash, we do not use discrete (binary) thresholds; instead, we use weighting coefficients that correlate with brightness and saturation. But before assessing the contribution of a flash, it first needs to be identified in the diamond image, and this is where the problem of defining what constitutes a colored flash and its boundary arises (some metrics do not have this problem as they simply count colored pixels, but in that case, they cannot account for the linear size of the flash).

In this cut, there are very dark blue Fire flashes with a soft brightness gradient on the pavilion facets under the table. It is very difficult even for the eye to determine where the colored flash ends. Usually, such dark colored facets appear in zones of partial leakage, where, for example, red and green rays exit through the pavilion, while the blue rays, due to their higher refractive index, partially remain in the diamond. Most Fire flashes are created by the eye cutting off part of the spectrum from a conical beam exiting the diamond, typically resulting in bright Fire flashes.

A more detailed analysis may take several weeks. However, it is already clear that a 0.1-degree difference does not create a significant difference in Fire for these cuts. The 0.1-degree difference highlights the rounding error in determining what is considered a Fire flash.

Whatever the criterion or threshold for what is considered a Fire flash, it is always possible to find a cut that creates flashes with this borderline threshold. Consequently, the slightest change in the cut will move such flashes from one category to another

Dear Sergey,

Your detailed explanation is of great value, I do appreciate your valuable time and consideration.

So this is what I conclude from the amazing information you shared:

although you are not using a binary threshold, you defined some boundaries in the range of the used coefficient; so when we get to these boundaries the scores may change suddenly, Am I right?

and also I don't understand exactly what kind of categories you defined for Fire for example, and why there should be such categories, is this related to the brightness threshold of a Fire flash? (This question is important to me, please)

I would really like to know about the correlation of brightness and fire (Linear, nonlinear, the coefficient numerical value?)



and I will study any information that you may share in precise detail, so no matter how complicated or long the story is I can continue to learn from you and discuss this for years, so exact scientific complicated data is what I'm looking for, Please.

please kindly consider, I'm still experimenting in the middle of unknowns so I can just share things when I understand the accuracy and reason for them. (I don't like to overplay my hand)
So sharing data and knowledge related to claims is inevitable but not in a rash manner. (Just ask me for an instant sharing if it matters to you)

your precious supports are enormously appreciated.

To all the friends that are following, I'll write my status and expectations in "red" from now on so the red part of my message is mainly written to the responders to build their mindset about what I'm saying, you may ignore the red part if you are not the responder of the message.

 

[Serg]

and also I don't understand exactly what kind of categories you defined for Fire for example, and why there should be such categories, is this related to the brightness threshold of a Fire flash? (This question is important to me, please)

I did not introduce categories for Fire, nor do we divide the Fire score into categories. I mentioned that there are two fundamentally different mechanisms for the occurrence of colored flashes. They differ in the stage at which part of the spectrum was cut off. When part of the spectrum is cut off due to leakage, the Fire flashes are generally much darker compared to when the spectrum is cut off at the stage of passing through the observer's pupil. However, even when the spectrum is cut off by the observer's pupil, very dark Fire flashes can still be observed.

The threshold for Fire is not determined by brightness but by chroma and the size of the flash. Chroma, in a simplified sense, is the product of brightness and saturation. If the flash is either very small, dim, or unsaturated, the observer will not recognize it as a Fire flash.

Imagine someone spilled ketchup or red wine on your wool trousers, and you had to send them to the dry cleaners. If the dry cleaners did a good job, you wouldn't see the stain. If you see the stain, you would consider the trousers dirty and think the dry cleaners did a poor job, regardless of how well you can see the stain or how contrasting it is. What matters is whether you see it or think you see it. There is no clear threshold for when a stain is visible or invisible after dry cleaning; its visibility depends on who is looking, whether they know where the stain was, the size of the stain, the pattern of the trousers, the color of the trousers, and the lighting. One person might see the stain and consider the trousers dirty, while another might not see it and consider them clean.

How do you determine the threshold of stain contrast at which trousers are clean? You could develop an objective device to measure the stain and calibrate it so that 99.999% of people don't see the stain, but then the trousers would suffer significantly from such cleaning. This again is a question of balancing between False Positive and False Negative errors. Demanding that both be zero is unreasonable for most real tasks. There is a reasonable level of errors, and those who demand that there be no errors, such as no errors in evaluating clarity or color, simply do not understand how much it would cost to halve the error rate. From a certain level, reducing the error rate by half requires increasing the costs by an order of magnitude. Sometimes it is worth it, sometimes it is not.

 

[Garry H (Cut Nut)]

I did not introduce categories for Fire, nor do we divide the Fire score into categories. I mentioned that there are two fundamentally different mechanisms for the occurrence of colored flashes. They differ in the stage at which part of the spectrum was cut off. When part of the spectrum is cut off due to leakage, the Fire flashes are generally much darker compared to when the spectrum is cut off at the stage of passing through the observer's pupil. However, even when the spectrum is cut off by the observer's pupil, very dark Fire flashes can still be observed.

The threshold for Fire is not determined by brightness but by chroma and the size of the flash. Chroma, in a simplified sense, is the product of brightness and saturation. If the flash is either very small, dim, or unsaturated, the observer will not recognize it as a Fire flash.

Imagine someone spilled ketchup or red wine on your wool trousers, and you had to send them to the dry cleaners. If the dry cleaners did a good job, you wouldn't see the stain. If you see the stain, you would consider the trousers dirty and think the dry cleaners did a poor job, regardless of how well you can see the stain or how contrasting it is. What matters is whether you see it or think you see it. There is no clear threshold for when a stain is visible or invisible after dry cleaning; its visibility depends on who is looking, whether they know where the stain was, the size of the stain, the pattern of the trousers, the color of the trousers, and the lighting. One person might see the stain and consider the trousers dirty, while another might not see it and consider them clean.

How do you determine the threshold of stain contrast at which trousers are clean? You could develop an objective device to measure the stain and calibrate it so that 99.999% of people don't see the stain, but then the trousers would suffer significantly from such cleaning. This again is a question of balancing between False Positive and False Negative errors. Demanding that both be zero is unreasonable for most real tasks. There is a reasonable level of errors, and those who demand that there be no errors, such as no errors in evaluating clarity or color, simply do not understand how much it would cost to halve the error rate. From a certain level, reducing the error rate by half requires increasing the costs by an order of magnitude. Sometimes it is worth it, sometimes it is not.

Thank you Sergey for excellent clear and comprehensible explanations.
1. I now understand why there are more blue and violet flashes than red and orange. (Green is often absent of course because the colours either side combine in the pupil to create white).
2. It also helps me understand one reason you prefer slightly deeper proportions them me. Personally the fact that diamonds are dirty 99% of the time (i.e. forever -1%) after they have been cut and polished, means that the pavilion leakage is enhanced. I respectfully ask you again to consider a study where the pavilion has a lower RI oil interface and see what a realistic result is for worn diamonds?

You stain example is well written.
I would extend it to the difference between people who have learned what to look for in a diamond, and hence are fussy about the appearance, and those that never really knew about the stains position / the intrici's of what we see when looking at a diamond.

 

[Garry H (Cut Nut)]

For anyone following this thread who is not aware of this article, please use as bed time reading.
Actually, jokes aside, when we wrote this article we expected it might take a decade + for it to be useful. It is now + a decade:

https://files.octonus.com/documents/GAA_Journal_V25_No3_Sept2013_V5.pdf

 

[Serg]

I respectfully ask you again to consider a study where the pavilion has a lower RI oil interface and see what a realistic result is for worn diamonds?

If the pavilion facets have a refractive index of 1.50 and the crown facets 2.42, the result will be roughly the same as for quartz. This happens because, for a round cut, the result is primarily influenced by the critical angle of internal reflection on the pavilion facets. And how will this result, whatever it may be, justify the expenses on R&D? Who would use parameters optimal for quartz for a round diamond? In the first approximation, you can get these results in both Diamcalc and Gemcad. I don't want to waste my time and the time of highly qualified and expensive programmers to create an exact ray path calculation for a case where the pavilion of a diamond is covered with oil, as the result is clear to me. Anyone interested in such fixed ideas should invest in testing them themselves.

I engage in R&D only if I understand how to use the result in case of success. Here, I don’t know how to use the result, whether in case of success or failure. Moreover, I don't understand what would even be considered a success here.