HCA questions

[searcher]

I came across this in an old thread. Is it really the development chart for the HCA? If so, has there been any further research into the the upper-center region? Why does a diamond with a 36 degree crown angle get a better score as the pavilion angle increases? Why is there a blue island around the 7.2 and the 7.6? Where can I find the published paper? I think the HCA is a useful tool and I am excited to learn more about the methods behind it.

 

[pricescope]

I forwarded this question to Garry. He didn't have internet access yesterday...

 

[Garry H (Cut Nut)]

there have been a few minor changes to that chart.
All the background is not hidden - it has always been available at www.diamond-cut.com.au

For some reason there are some nice looking stones around the 36 crown anglre and steep pavilions.
the other blue zone has strong table leakage.

 

[searcher]

Thanks for the reply. I've spent the last few hours reading your site and the material at http://www.gemology.ru/cut/index.htm. Good stuff.

This statement is tough as nails:

"So be it if the methods and results of this cut study are unscientific, if the results are wrong then I will enjoy the search for better solutions. Criticism is a friend, not a foe in the quest for continual improvement. "

In this spirit I have a few questions:

Are the HCA results based solely on computer models or have you conducted some more studies with real diamonds?

You mention that scientists have confirmed parts of your research. Any references?

You state that you assume jewelry store lighting with illumination on the crown only http://www.diamond-cut.com.au/10_method.htm Is the difference with the MSU study that they did theirs under diffused "office" lighting? Are you working on an HCA for diffused lighting?

What are the most controversial parts of your research? What are the major points of agreement?

Thanks in advance and for caring about improving the quality of diamonds. I like to consider myself a scientist, but I don't have any experience in the diamond industry other than buying an engagement ring. I guess it bothers me a little that no one on this forum is asking any questions other than how do I use this thing. Do any appraisers/cutters/vendors/anyone else want to chime in?

 

[valeria101]

----------------
On 6/19/2004 2:54:58 AM searcher wrote:

I guess it bothers me a little that no one on this forum is asking any questions other than how do I use this thing.

----------------

Oh, but there was allot of talk about the HCA... both good and bad (you can track down the threads among Garry's or Jonathan's posts rather than a generic search). The topics were mostly on how far one can go using the HCA and how the results relate to either other tools and visual perception (based on empirical results). Other notes are posted permanently as part of "tutorials" here and there (Gemappraisers, Niceice, DBOF). The topic is definitely not new.

I do not remember to have read much about what difference diffuse lighting would make: you may find this issue debated alongside contrast brilliance (mostly at GOG).

To what extent ray trace modeling approximates these "diffuse lighting" is yet another Q which I never heard of here (a comparison between the IdealScope and FireScope posted by GOG touches the issue indirectly - no model constructed, just examples); Jeff Graham used to have some comments posted on his site. In theory, there are alternative models but I never saw anything but ray trace analysis used in relation with gem faceting.

 

[searcher]

On 6/19/2004 1:38:03 PM valeria101 wrote:











Oh, but there was allot of talk about the HCA.

I checked out most of it. Usually these are arguments about the usefulness of the HCA.

To what extent ray trace modeling approximates these 'diffuse lighting' is yet another Q which I never heard of here (a comparison between the IdealScope and FireScope posted by GOG touches the issue indirectly - no model constructed, just examples); Jeff Graham used to have some comments posted on his site. In theory, there are alternative models but I never saw anything but ray trace analysis used in relation with gem faceting.

Tolkowsky (1919) picked the five most lively diamonds he could find and confirmed that they fit his model. http://www.gemology.ru/cut/english/tolkow/tolk11.htm In my opinion this is the main reason his work has stood up so well over time. For all I know he could have thrown out his models of BIC's and FIC's because the actual diamonds that had these proportions looked ugly to everyone he asked.

I would like to see a double-blinded study using real diamonds and the human eye. Where is it?

 

[strmrdr]

----------------
On 6/20/2004 9:50:35 AM searcher wrote:

I would like to see a double-blinded study using real diamonds and the human eye. Where is it?

Gia is finishing up on one right now.
How much of the information they will make available is not known.

 

[strmrdr]

As far as the hca goes Im still undecided on it myself.
It appears useful and seems to match a lot of what the experts are finding out by other means.
But the biggest problem with it is that it doesn’t take into account all the facets which could have a huge impact on performance.
Also it uses averages which can be deceptive.

The hca combined with h&a photos and idealscope/firescope images and the eye's of the vendors recommended here will get you into the top 1% of diamonds being sold today.
Trying to sort out that 1% is the real problem.


Now as far as computer ray tracing no it is not all that accurate a diamond just has too many interaction points for a simple model to work well.
With enough time and computing power a very good model could be developed but I am highly suspicious of a model that gives results in under a second on a desktop pc.
There is just too much going on inside a diamond combine that with the huge variation in behavior of light which is not fully understood in itself and it becomes hugely complicated.

 

[Garry H (Cut Nut)]

Strm DiamCalc is not simple ray tracing it is complex beam tracing and takes about 30 seconds to calculate light return and contrast.

 

[strmrdr]

cut nut: it is still a simplified model.
GIA couldnt get a full trace taking all 57 facets and all the facet junctions into account using super-computers it isnt going to happen on the desktop anytime soon.

 

[strmrdr]

This diamond:
diamond

ideal scope image
here:

 

[strmrdr]

The blue dots are 440nm laser light at .5W
The red dots are 685nm laser light at .5W
All are aimed at a point 1/2 way up the pavilian of the diamond to hit a facet junction under them and are 1mm wide when they hit the diamond.

At what points would red,blue and magenta light be returned from the diamond?

Now reverse them and compare the difference in light return points and plot the difference between the return points and note the affects of wavelenth on angle of return.

Then aim them at different points on the pavilian and repeat the test.
Repeat the test aiming for a facet junction opposite where the beam enters instead of the one under it.

Do that one on a desktop :}

 

[Garry H (Cut Nut)]

I am not sure what the point is Strm?

Do you have a copy of DiamCalc?
It does not base its calculations on ray's of single colors (i think you would call that reverse ray tracing).

GIA and super computers with software designed by GeoChemists has little to do with the work of russian physicsts and mathematicians. You may remeber that GIA did not understand polarisation and how to calculate its effects in their explanation of their approach on the website a couple of years ago (they retracted and re edited it).

At this time the bigger question is how can you build a 3D model of a diamond that is acurate enough to perform beam tracing analysis?

AGS began working with Dr Jose Sassian - prof Optonics from the Arizona Uni last year and were using various ray tracing software packages. At that time AGS had a very old version of DiamCalc. Peter Yantzer saw some of their proposed lighting models on DiamCalc at Vegas last year and now AGS are using DiamCalc almost exclusively in their research program for the new round and fancy shape cut grades they hope to release next year.

 

[valeria101]

"Beam tracing" is a new word to me. Allowing a set of properties to a "ray" sounds like a logical extension of simple ray trace.

Maybe this is the right thread to ask... why aren't more diamond representations following the recipe of H&A?

The idea to represent a pattern of visually significant "patches" (contours, facet reflections, whatever useful to define) sounds interesting...

and practical to save computing power... and good to produce appealing (as well as informative) visual representations of diamonds. After all, it takes a leap of faith to even consider the mangled patches of the Iscope picture of a non ideal stone regardless of whether the preference for geometrical order has any meaning.

There are many faceting patterns made to contain appealing symbols - so the idea seems natural, even if only the H&A got picked up to date.


For example, the representation presented by the Isee2 interface. I know only what shows on the screen, but to reach that end the model seems to take into account some deliberate visual unit: sections limited by the reflection of facets onto one another. To me, this seems a handy analytic tool.

The same type of "unit" could be implicit to quantitative representations
(such as the photo realistic images) - getting progressively broken down as layers of details are added. Likewise, lighting conditions and technical specifications of 'scopes and viewers would vary the size of this observation unit - much as happens with seeing gems unaided while being worn.

Is this assuming too much ?

 

[valeria101]

----------------
On 6/20/2004 9:50:35 AM searcher wrote:

I checked out most of it. Usually these are arguments about the usefulness of the HCA.

----------------[/quote]




Definitely not all - and this is why I pointed to seller's and experts' posts, not shopping advice. You will find a good doze of more or less friendly criticism (AGA, DBOF...) and counterexamples (posts by MDX and NiceIce come to mind). Besides, the HCA is supposed to be a screening tool and more precise screening for cut quality is available.

 

[searcher]

----------------
On 6/20/2004 5:23:58 PM Garry H (Cut Nut) wrote:








At this time the bigger question is how can you build a 3D model of a diamond that is acurate enough to perform beam tracing analysis?
----------------




If you're this unsure about your model, then why would you recommend nontraditional diamond proportions such as the BIC and FIC? Is it safe to monkey with Tolkowsky proportions before we know the effect it may have on all facets, and before we know that the resulting diamonds are actually pretty to look at under a variety of lighting conditions? Wouldn't a cut-conscious buyer be safer to avoid the BIC and FIC until this model has been properly tested?

 

[valeria101]

If you're this unsure about your model, then why would you recommend nontraditional diamond proportions such as the BIC and FIC? Is it safe to monkey with Tolkowsky proportions before we know the effect it may have on all facets, and before we know that the resulting diamonds are actually pretty to look at under a variety of lighting conditions? Wouldn't a cut-conscious buyer be safer to avoid the BIC and FIC until this model has been properly tested?
----------------[/quote]




There are good reasons to believe that these models have been at least more tested than the theory that went into Tolkowsky's paper. Surely you are aware that some recommendations of the initial study remained forever impractical ("knife edge girdle" and the pinpoint set of proportions). "TIC" does not mean dead on Tolkowshy study or following more precise research. Aren't "BIC", "TIC" and "FIC" all mentioned as separate categories on the HCA? And so is the AGS0 ideal range. The relation between HCA scores and most other definitions of "ideal cut" are also cited on preciousmetals.com.au.

The problem... there is no consensus (yet, at least) on what makes a RBC ideal. If more industry backing is a good answer for you, you can always wait for GIA's opinion when their cut grading will get ready.

Until then, it doesn't sound too bad to have a range of opinions and research results to judge. Anything wrong with it?

 

[strmrdr]

----------------
On 6/20/2004 5:23:58 PM Garry H (Cut Nut) wrote:

I am not sure what the point is Strm?


Do you have a copy of DiamCalc?

It does not base its calculations on ray's of single colors (i think you would call that reverse ray tracing).


GIA and super computers with software designed by GeoChemists has little to do with the work of russian physicsts and mathematicians. You may remeber that GIA did not understand polarisation and how to calculate its effects in their explanation of their approach on the website a couple of years ago (they retracted and re edited it).


At this time the bigger question is how can you build a 3D model of a diamond that is acurate enough to perform beam tracing analysis?


AGS began working with Dr Jose Sassian - prof Optonics from the Arizona Uni last year and were using various ray tracing software packages. At that time AGS had a very old version of DiamCalc. Peter Yantzer saw some of their proposed lighting models on DiamCalc at Vegas last year and now AGS are using DiamCalc almost exclusively in their research program for the new round and fancy shape cut grades they hope to release next year.


----------------

Iv checked out the demo version I cant afford the full version of diamcalc.
The point is that I presented a non-simplified model to work with and one that would not be done on a desktop machine in 30 seconds.
2nd no one seems to have addressed the affects of wavelenth on diamond light return and I was hopeing that someone would point a study out or it would prompt someone to do one :}
Im not trying to downplay the work of the creaters of diamcalc and the studies behind it but to claim that you can learn all that is needed to be learned by using it is a real stretch.
Any program is only as good as the model behind it for example how many rays does it use in its calculations, real world there would be millions interacting at once.
For example 2 rays may get split into different colors at some point but returm out the same point of the diamond so the resulting color would be a combination of the 2.
It may also be washed out by the surface reflection and never seen.
Also my example hit on another problem area it is fairly easy to calculate return paths of beams that strike the surfaces of the reflectors but what of those that hit the junction points of the facets?
From my experence with lasers Im guessing that it will throw an element of randomness into the equation that would be almost impossible for a computer program to predict.

 

[Garry H (Cut Nut)]

DiamCalc calculates and models light paths until they become nelgigable (I do not remeber the %).
Facet junctions on diamonds are sharp not rounded (rounded is a nearly impossible concept on diamonds) so diamonds do not have non sharp edges.

If you work in this area then $280 is not a large expense.

DiamCalc does all the things you mentioned and a lot more. Perhaps you are a shill?

Asking such questions so that I place a free advertisment for DC?

 

[Garry H (Cut Nut)]

http://www.cutstudy.com/cut/english/optical/index.htm
the AGS research before updating Diamcalc.
http://www.cutstudy.com/cut/english/grading1/index.htm
there are about 16 pages here that explain the problems with ray tracing / parallel beams and many other factors.
It is older than the current DiamCalc system.

Strangely Searcher the idea that 3D software does not work because of computing power is wrong - more troublesome is matching human pyshopysiological responses than the actual number crunching.

 

[strmrdr]

$280 is a lot of money, Im a diamond junky computer geek im not in the business.
Im spending less than that on my wedding ring lol.
Im also cheap :}

As far as 3d modeling being wrong due to computing power again your reading more into it than I said.
It may be dead on as far it goes but how far does it go is the question.
The time it takes to crunch the numbers is a real word factor in all 3d modeling and a diamond is not an easy thing to model nor trace.
There for almost always a simplified model is used and the choice of that model and how many traces are run determines the accuracy of the results.
At 30 seconds on a desktop machine they are not using millions of traces and a very complex model.
They are either doing few traces on a complicated model or a lot of traces on a simple model.
That is just a fact of computing power.
Its entirely possible that it is actually the best diamond modeling software available right now but that doesn’t change the way things work in the real world.

Then you have the major issue that you touched on is how much of this actually has real world relevance

 

[Garry H (Cut Nut)]

Storm you are really out of your depth.
Spend some time on the cutstudy.com site and learn more about it all before you make these desparaging and uninformed comments. You are wrong. The number of traces depends on the amount of light you need to follow. in most diamonds each ray traces around 5 reflections until less than 1% of the light remains. There is no need to go beyond this. The number of rays is more than adequate to model what humans can percieve.
Either buy it and see for yourself or do the research for free on the website. Then come back with informed comments rather than non useful conjecture.

 

[searcher]

----------------
On 6/20/2004 10:59:06 PM valeria101 wrote:











There are good reasons to believe that these models have been at least more tested than the theory that went into Tolkowsky's paper.

Reference?

Surely you are aware that some recommendations of the initial study remained forever impractical ('knife edge girdle' and the pinpoint set of proportions). 'TIC' does not mean dead on Tolkowshy study or following more precise research.

My point was that he made some effort at the end of his study to tie his results in with real world observations.

The problem... there is no consensus (yet, at least) on what makes a RBC ideal. If more industry backing is a good answer for you, you can always wait for GIA's opinion when their cut grading will get ready.

Until then, it doesn't sound too bad to have a range of opinions and research results to judge. Anything wrong with it?

I'd like to understand the tools I'm using. Anything wrong with that?

 

[searcher]

----------------
On 6/21/2004 1:15:50 AM Garry H (Cut Nut) wrote:











Strangely Searcher the idea that 3D software does not work because of computing power is wrong - more troublesome is matching human pyshopysiological responses than the actual number crunching.

---------------

Like I could have a protein in my retina that varies from yours by one amino acid. Therefore there is no way of knowing that if we describe a certain wavelength as the color green that we're really seeing the exact same thing. Maybe someday we will be able to pick diamonds for individuals based on their genotype.

Back to the original graph, do you have any speculation as to why it is shaped the way it is? I find it really weird but not impossible that a computer model could come up with a strange island and a peninsula like that. What am I missing? If the blue zone has strong table leakage, why doesn't the zone above it have even worse table leakage? Why are we so concerned about light leakage if we don't know for sure that it looks bad? A mirror would leak no light at all. Why did you distribute the data points as you did?

By the way, what's up with your response to storm? You should be able to explain your methods without making someone learn how to use computer software. If you can't talk to us, I don't think you stand a chance among your peers.

 

[valeria101]

----------------
On 6/21/2004 6:26:35 PM searcher wrote:

----------------
On 6/20/2004 10:59:06 PM valeria101 wrote:

Reference?

I'd like to understand the tools I'm using. Anything wrong with that?

----------------

Nope. Since PS doesn't publish a scientific journal, I can only direct you to less formal sources: the database and notes at Good old Gold and the OctoNus site. I would think you already went through GIA's study. You will find some mention of panel studies on the perception of brilliance and so forth. I am not aware of how much of the background research has ever been made available by OctoNus and peers.

On the other side, Jonathan at GOG has developed (and posts) his own empirical research. If you find any other publicly available data of this sort, let me know

Many times there has been mentioned (see the links I cited previously) that the worst thing about the HCA is the temptation to over interpret it's output

There is no shortage of precise analytic tools. The HCA remains the easiest, costless tool to make some sense of diamond stats. I think "easiest" and "costless" are important attributes here. Once one holds the diamond, much better precision is easily feasible but with only four numbers to go by any analysis gets uneasy.

So... if what you need is to understand this system, you may want to factor in data availability and cost as constraints. Otherwise, there are bigger, better and more expensive toys to play with