Diamond Cut Research Technical FAQ
Approach
Why are so many appearance and quality aspects needed to assess the overall cut quality of a round brilliant cut diamond? Why include brightness, fire, scintillation, polish, symmetry, durability, and weighting factors as separate aspects of this system?
Why does the GIA System take the final grade from the attribute with the lowest value, instead of averaging the values of the different aspects of a diamond’s appearance?
Why does the GIA System have five grades?
Why doesn’t the GIA System have a separate, explicit scintillation metric?
If clarity characteristics and/or body color can affect light return, why are they not included in the GIA Diamond Cut Grading System?
Observations
How was the observer-diamond geometry determined for the observation tests and modeling used to develop the GIA Diamond Cut Grading System? In your modeling, do you account for the rocking and tilting of diamonds during observation?
How were brightness and fire metrics determined using the observation data? Were these data sets too scattered to be useful?
[ top ]
Modeling
Do your metrics and observer viewing conditions incorporate aspects of both trade and consumer environments?
I’ve read about “Head Shadow”—is it included in GIA’s computer modeling? If not, why not?
The dark area in the GIA brightness metric is 23 degrees in radius; this seems too large for a “head shadow”. Doesn’t this make many diamonds appear dark?
What about stereo vision? Does the GIA model include it? And what about pupil size?
When you used domes for establishing the brightness metric, did the observers close one eye to look through the holes in the domes? Wouldn’t this be a problem since humans use two eyes?
The GIA fire metric is modeled using a single point light source; however, there are lots of lights in modern jewelry stores (e.g., multiple light sources in display windows, rows of spot lights above or within jewelry cases plus fluorescent tubes on the ceiling), and many lights in the GIA common viewing environment. Doesn’t this difference create a discrepancy?
[ top ]
Measurements
Why does GIA measure star and lower half facets?
Why are different measurements reported for the same diamond samples in the Fall 2001 and Fall 2004 Gems & Gemology articles?
Why doesn’t GIA report measurements to greater precision?
Is it realistic to expect diamond manufacturers to cut to these sets of proportions?
[ top ]
Results
Will there be diamonds in GIA’s highest category that don’t look good?
Does a diamond with a higher cut grade always look better than one with a lower cut grade?
Did GIA confirm that Tolkowsky’s cut proportions give the “best” grade? Or are there sets of proportions that look as good or better than Tolkowsky’s?
If a diamond looks better in one of the proprietary viewers (such as one of the “hearts and arrows” viewers), does it get a better grade in the GIA System? Did observation tests confirm that such diamonds look better?
How does “optical symmetry” relate to the symmetry that GIA currently reports, one aspect of finish?
What is the effect of the mounting and the background against which the diamond is viewed on the appearance of a round brilliant cut diamond?
[ top ]
Approach
Why are so many appearance and quality aspects needed to assess the overall cut quality of a round brilliant cut diamond? Why include brightness, fire, scintillation, polish, symmetry, durability, and weighting factors as separate aspects of this system?
From the many conversations we had while carrying out our observation tests, we found that all these aspects need to be considered when assessing the overall cut appearance and quality of a round brilliant diamond. A diamond is attractive when it is bright and fiery; when it has a pleasing, even pattern to its areas of light and dark (scintillation); and when it demonstrates a high level of craftsmanship. A diamond is less attractive if it is dark, if it is not fiery, if it has a distracting or unpleasant pattern, if it is carelessly polished, or if it is sloppily cut. Also, a diamond with a too-thin girdle may be attractive, but it is more susceptible to damage. A diamond with a thicker girdle or a larger total depth also can be attractive, but then the customer pays for unnecessary weight (this is illustrated in the article). So to consider “overall cut appearance and quality,” we needed to address these concerns as well.
Therefore, this system is highly detailed internally, but it yields a single, overall grade for cut quality. We will be providing online and standalone tools for members of the trade to easily look up the grades of their diamonds when they have the appropriate measurements on hand. We are sensitive to the industry’s needs and wish to keep our grading system simple and straightforward.
[ top ]
Why does the GIA System take the final grade from the attribute with the lowest value, instead of averaging the values of the different aspects of a diamond’s appearance?
Through observation testing, we found that each factor was important, not just the average of the several factors. So we arrived at a grading method that uses neither averaging nor “weighting”: instead, a diamond must score well in every aspect in order to get a high grade. This method produces results that most consistently agree with our results for overall appearance, and with the opinions learned from our extended interaction with the trade and consumers.
[ top ]
Why does the GIA System have five grades?
In a word, practicality. This number of grades is derived from extensive testing as to how finely an experienced observer can reproduce results within “standard tolerance”. For instance, one of the exercises we asked trade observers to do was to rank a group of eight to fifteen diamonds into however many different grades they chose. This exercise demonstrated to us that less than five grades was inadequate, but that empirically, most observers would not easily see the same differences as each other with more than five grades.
[ top ]
Why doesn’t the GIA System have a separate, explicit scintillation metric?
Instead of one modeled and calculated metric, we approached scintillation differently. If scintillation is considered as a pleasing pattern of light flashes that seem to sparkle as the diamond moves, then it can be broken down into three aspects: pattern, bright flashes, and movement. We first concentrated on pattern, as the bright flashes were already included in the brightness and fire metrics.
From conversations with tradespeople as they examined and compared diamonds in one or more environments, we learned that appearance patterns were frequently described by negatives, by what was wrong with them—e.g., “too dark under the table,” or “the pattern goes away when you tilt the diamond slightly.” We studied which aspects of diamond geometry produced these and other (both positive and negative) patterns. We analyzed the different pattern aspects in relation to various sets of proportions, and found consistent relationships between these proportion sets and the observations. A system of assessing scintillation based on these proportion sets eliminated the need for a separate scintillation metric.
Historically, scintillation has been linked to movement. We learned, however, that observers often focused on certain positive or negative patterns within the diamond regardless of movement. Individual observers moved the diamonds in different ways, and to differing degrees, yet they produced a coherent set of observations with regard to pattern factors. Once we had successfully related these pattern-observation results to proportion combinations (that is, such that predictions matched later observations), we found no reason to add a separate factor for movement.
[ top ]
If clarity characteristics and/or body color can affect light return, why are they not included in the GIA Diamond Cut Grading System?
Our observer tests enabled us to examine the effects of other diamond quality factors (such as these) on overall cut appearance. Although with diamonds of very low color and clarity we found some impact on overall appearance, in general, observers were able to separate these factors out of diamond assessments (that is, we found that differences in cut quality were reliably predicted and observed when diamonds had similar color or clarity grades, even in the lower grade ranges, and observers felt comfortable assessing cut quality for diamonds of very different color and clarity grades when they had prior knowledge of those grades for the diamonds they were observing).
The complete “picture” of a diamond is achieved by understanding the appearance effects of all the factors on its grading report, which includes color and clarity. Even though the overall appearance is a combination of these factors, each is graded independently. Since the GIA Gem Laboratory already provides the color and clarity of a diamond on its grading report, we considered that it could be confusing (if not unfair) to include the effects of these characteristics (which are not directly related to cut) in the overall cut grade as well.
[ top ]
Observations
How was the observer-diamond geometry determined for the observation tests and modeling used to develop the GIA Diamond Cut Grading System? In your modeling, do you account for the rocking and tilting of diamonds during observation?
In the initial stages of our observation research, we visited a number of experienced diamond manufacturers, brokers, and retailers, and measured physical relationships in the environments they used to observe diamonds. These measurements included ranges of diamond-to-light-source and diamond-to-observer distances, ranges of viewing angles (tilts), and types of illumination and their intensities.
The common viewing environment we developed and used, and the method for observing diamonds in this environment, define the most important common elements of these observer/observation factors that we found to be critical to consistent and reliable observations of diamond appearance. For example, we used our findings on observer-diamond distance to determine the viewing range that provided the most consistent results.
Our common viewing environment consists of a neutral gray box with a combination of daylight-equivalent fluorescent bulbs and overhead white LEDs (light-emitting diodes). Intensities of the two light sources were established by determining when a set of reference diamonds showed the same relative amounts of brightness and fire respectively as they showed in the dealer- and retail-equivalent lighting conditions we described in the article. In this way we were able to combine the observable appearance aspects of brightness and fire in a single viewing environment, while also preserving the general qualities of both dealer and retail lighting.
Our modeling conditions are quite different. For both our brightness and fire metrics, the light source and the observer are at infinite distance; these choices simplify some of the mathematics used for modeling. Through an iterative process, we verified that the modeling results (using these theoretical geometries) matched human observations of real diamonds made with domes of finite size and real light sources at measurable distances; at the same time we refined some of our modeling conditions, as described in the article.
The tilt of a diamond (relative to the observer), and the observer’s rocking of a diamond while examining it, are accounted for in several ways in our observations and modeling. The tilt is explicitly incorporated in our fire metric in that we consider all observer viewing angles, weighting our results to emphasize those that are most important (i.e., those located closest to the “table up” position). In addition, our assessment of pattern-related scintillation takes the effects of tilts into account in that those diamond proportion sets whose patterns become less attractive when rocked or tilted are downgraded in our system. Finally, the effects of different tilts are contained within our system through our process of matching predicted results to actual observations of diamonds in typical environments; since observers rocked the diamonds as they were observing them, the effects of this factor were automatically incorporated into our grading criteria.
[ top ]
How were brightness and fire metrics determined using the observation data? Were these data sets too scattered to be useful?
Preference data are almost always scattered; but if one gathers enough observations, powerful mathematical tools are available to analyze them. Using such mathematical tools, we chose the metrics that best fit our data. We felt that the grading system should account for different opinions about appearances expressed by various observers (that is, two different appearances could be in the same category if there was no general preference for one over the other). When we applied statistical analysis, we found that we could not only evaluate the varied responses we obtained, but we could also build a grading system to account for these varied responses. Out of this we determined that we could predict a range of appearances that most observers would agree belonged in certain categories or rankings.
However, we also found that, as diamond dealers told us, a person should look at many diamonds and buy the diamond that they themselves prefer. While we can rank diamonds based on what is preferred most often, purchasers should choose each diamond based on their own visual preferences.
[ top ]
Modeling
Do your metrics and observer viewing conditions incorporate aspects of both trade and consumer environments?
Consumer environments can vary quite a bit, as consumers typically buy their diamond jewelry in a store, and then wear it in nearly every environment. While diffuse lighting is often typical for most offices, there are a variety of mixed lighting environments that consumers work and live in. Although we felt it was important to develop a viewing environment that was consistent with the places in which consumers would view their jewelry, we believed it was critical that the environment be logical for those who regularly make essential diamond manufacturing and purchasing decisions.
We had observers look at diamonds in a variety of environments. We found that observers made the most consistent observations in diffused light, and that these observations were remarkably similar to those made in environments used by most diamond dealers (in which a fluorescent overhead source was always used). As spot lighting (often present in many retail environments and other locations where consumers might wear diamond jewelry) was introduced into that environment, consistency was maintained up to a certain threshold. After that point, as spots increased in number or intensity, the consistency of comparisons dropped.
Although we use very specific lighting assumptions in our metrics to emphasize the particular appearance aspect we are interested in (diffuse lighting for our brightness metric and spot lighting for our fire metric), we combined these lighting types in our observation environment used for overall appearance so that all factors of diamond appearance could be assessed.
The decision to use a combination of lights was further supported when we asked retailers to make judgments in their own stores as if they were buying. While a few included some type of spot lighting to see the sparkle and fire, nearly all made their judgments, and especially final evaluations, in diffused lighting.
[ top ]
I’ve read about “Head Shadow”—is it included in GIA’s computer modeling? If not, why not?
“Head shadowing,” the effect of the head of the viewer leaning over the diamond and disrupting the lighting, is only one aspect of the potential influence of the observer. When looking at the diamond from a typical distance, the observer is part of the panorama, but is neither the brightest part (the light source is) nor the darkest part (the empty space behind him or her may be darker, as was the case in many diamond dealer environments). The combined effects of all light and dark areas are considered in our brightness metric. We also took shadowing by the observer into account when we determined scintillation factors that describe negative pattern effects.
Most references we have seen to head shadow take the width of the observer’s head and use it to define a round circle obstructing the light source. In fact, the head is oblong and it rarely appears directly between the light source and the diamond. We also have found that the whole upper body creates an influence on the diamond’s appearance. These factors further support a metric that addresses the combined effect of all light and dark areas.
[ top ]
The dark area in the GIA brightness metric is 23 degrees in radius; this seems too large for a “head shadow.” Doesn’t this make many diamonds appear dark?
The (top) dark area for our brightness metric is 23 degrees in radius, centered at the dome zenith; however, it does not represent a “head shadow.” Observations through hemispheres that re-created the metric environment were used to establish relative rank order for the appearance of brightness, not to match patterns observed in more typical lighting. The rank order for brightness determined from metric calculations using this environment was the best match to that determined by observers in our common viewing environment.
[ top ]
What about stereo vision? Does the GIA model include it? And what about pupil size?
One could also ask about light- or dark-adapted vision, edge detection, and visual adaptation, among other related topics. When we discussed these issues with specialists in the field of vision science, we learned that human vision is a subject of active research among biologists and psychologists. We also realized that even if we modeled human vision as well as possible, we still would not be able to model what an observer values in the appearance of a polished diamond. So instead we used human observers, and tested our metrics empirically to determine which factors might still be missing from our model. As we developed the metrics and other predictions into a comprehensive grading system, we continued testing it against human observations. This empirical approach allowed us to create a system with high predictive ability without a detailed model of human vision.
[ top ]
When you used domes for establishing the brightness metric, did the observers close one eye to look through the holes in the domes? Wouldn’t this be a problem since humans use two eyes?
First, the domes were not used to “establish” the brightness metric; they were used to check that our computer models gave the same results as human observers for the same environmental conditions. The final brightness model we adapted was the one that best fit the brightness observation data from human observers using both eyes, and using a standardized lighting environment based on actual environments used in the trade. By ranking with observations in real environments, we use the actual human visual system to evaluate brightness, rather than a computerized attempt to model the complex visual system.
Second, observers were not asked to close one eye. In fact, as part of the set of dome observations, some large domes were built with holes up to three inches in diameter, in which observers could look with both eyes; these gave the same observation results as smaller domes with the same configuration. These tests confirmed what a vision specialist told us—that when looking at most small objects, a dominant eye takes control.
[ top ]
The GIA fire metric is modeled using a single point light source; however, there are lots of lights in modern jewelry stores (e.g., multiple light sources in display windows, rows of spot lights above or within jewelry cases plus fluorescent tubes on the ceiling), and many lights in the GIA common viewing environment. Doesn’t this difference create a discrepancy?
A model is a mathematical approximation of reality—not a duplication—that is useful for a particular application. For our research, the usefulness was judged by how well the model’s predictions matched observations of actual diamonds in these environments. In this case, the fire metric uses one light and many observer positions, while the observation data were gatheredusing many lights and single observers who rocked the stone while observing it. We used early observations to choose the best fire metric, and found that later observations matched the model’s predictions well.
[ top ]
Measurements
Why does GIA measure star and lower half facets?
Although they are sometimes called the “minor” facets, the stars, related upper-half facets, and lower-half facets cover about half the surface of a round brilliant, and several kinds of evidence indicate the importance of their influence on the overall appearance of a polished diamond. If we neglect the light rays that reflect off the surface (as glare) or that pass through the diamond (as leakage), most light rays interact with the round brilliant diamond’s surface seven times or more, including at least one interaction with a “minor” facet. In terms of energy, about three-quarters of the light traced through a typical virtual diamond interacts with at least one of these facets.
All three of these facet groups play a strong role in producing the face-up pattern exhibited by a round brilliant. Changing the lengths of these facets alone, independent of the other proportions, will change the diamond’s overall appearance. Our research has shown that metric calculations and human visual observations alike depend significantly on these two proportion parameters. In short, omitting these facets is like ignoring half the diamond.
[ top ]
Why are different measurements reported for the same diamond samples in the Fall 2001 and Fall 2004 Gems & Gemology articles?
Quoting the Fall 2004 Gems & Gemology article (p. 206): “Research Diamonds RD01-RD27 and RD 29 were previously reported in Reinitz et al. (2001); variations in proportion values from that article are the result of recutting, measuring device tolerances, and/or the application of rounding.” Since the 2001 article, we have been able to reexamine and refine our measuring practices based on advancements in measurement technology. In the most recent paper, the measurements we report have been rounded in a manner consistent with the standards of our grading system. As explained below, we believe that the new measurement values represent an appropriate precision given the ability of observers to consistently discern differences in appearance and of manufacturers to reproduce the specific proportions.
[ top ]
Why doesn’t GIA report measurements to greater precision?
There are three main aspects to this issue: how much precision is needed for our cut system (as indicated by the observation results), how much precision is practical for cutters to achieve, and the reproducibility of measurements (rather than the precision of any single measuring instrument). Although we could report measurements to a finer scale, we determined—based on the reproducible distinctions made by human observers—that such a high level of precision was not warranted. To check this, we examined each proportion parameter at two or more levels of precision to explore the differences in calculated metric values and the ultimate grade. Our results demonstrated that very subtle changes in measurements seldom altered a final prediction, or human assessment, of overall appearance; there was no need to demand tighter tolerances at this point.
In some cases (e.g., table size or pavilion angle), our reported measurements are the tightest that are practical, given how reproducible those measurements are on the same or a similar, properly calibrated, measuring device. In addition, we carefully considered the level of precision that would be sufficient for our system without over-burdening cutters. We then conducted experiments with actual diamonds to check and support this level of precision. By not demanding more precision than visual discernment supports, we feel we have reached the best practical, reproducible solution for consistent cutting and reporting.
[ top ]
Is it realistic to expect diamond manufacturers to cut to these sets of proportions?
Note that we do not have a single recommended set of proportions; instead, we predict appearance across a wide range of proportions. Note also that there are many ways to get a very high or high grade. This should provide diamond cutters with a multitude of possibilities.
However, our system does require that cutters be mindful of all facets, not just the table, bezels, and pavilion mains. For this reason, we spent time with cutters, evaluating how they responded to the various aspects of our new system: we needed to determine if this approach was realistic. We found that, although cutters may have to change some of their thinking, consideration of these additional parameters (such as star facet length) is not unrealistic.
[ top ]
Results
Will there be diamonds in GIA’s highest category that don’t look good?
For our observation testing, we included diamonds with many of the proportion sets that we predicted would be in the highest cut appearance and quality categories precisely to prevent any less attractive diamonds from being “overrated.” However, since each of the highest categories contains diamonds with a range of appearances, not every consumer will be equally attracted to every diamond in the same category.
In fact, we found that even our most seasoned trade members who claimed to follow traditional cutting styles did not always agree on which diamonds were best in the highest category. However, for each grade category, a majority of observers stated that they preferred the diamonds overall in that category to the diamonds in the next lower category. The disagreements we found could be attributed to differences in taste between observers. Personal opinions and cultural biases will undoubtedly influence the selection of a specific diamond.
[ top ]
Does a diamond with a higher cut grade always look better than one with a lower cut grade?
This is very much a preference issue. Some observers like certain patterns in diamonds, or a particular balance of appearance factors, that others do not. On this basis, one could prefer a diamond that earns a lower grade from appearance factors that are not so important to that particular person. However, all diamonds in a higher cut grade category will look better than diamonds in the next lower cut grade category to the majority of observers.
[ top ]
Did GIA confirm that Tolkowsky’s cut proportions give the “best” grade? Or are there sets of proportions that look as good or better than Tolkowsky’s?
Remember that the diamonds current in Tolkowsky’s time, and for which he made his calculations, “ideally” had knife-edge girdles, moderately large culets, and short lower-half facets. That said, a “modern” diamond with Tolkowsky’s proportions for most parameters but with a thicker girdle, longer lower half facets or halves, etc., would in most cases do very well in our system (as long as it was well-crafted). However, our system does not differentiate among diamonds in the same grade. Our research revealed that many combinations of proportions produce a diamond that deserves a top grade, and we documented many cases where proportions other than Tolkowsky’s were preferred by observers. However, the observations did not support any single set of proportions as “best.”
[ top ]
If a diamond looks better in one of the proprietary viewers (such as one of the “hearts and arrows” viewers), does it get a better grade in the GIA System? Did observation tests confirm that such diamonds look better?
Many people in the trade use the term “optical symmetry” in referring to “branded” diamonds that show near-perfect eight-fold symmetry by displaying eight “hearts” in the pavilion-up position, or eight “arrows” in the face-up position, when viewed in specially designed optical viewers. To investigate the possible benefits of optical symmetry, we included several such diamonds in our observation testing. We found that although many diamonds with optical symmetry received high observation scores, other diamonds (with very different proportions and, in many cases, no discernable optical symmetry) were ranked just as highly.
An interesting by-product of the testing was that those trade members who emphasized this type of diamond in their business generally chose such a diamond as the highest ranking (although not always). Those who did not market this type of diamond chose it as best about as often as they chose other diamonds we have placed in the top grade categories. It appears that these types of diamond could be likened to an “acquired taste” or “learned bias.” This doesn’t mean that some of them shouldn’t rank highly—it just means that not everyone agrees.
Finally, there are a variety of proportions that yield these patterns, some of which cause the diamond to appear darker to many observers. When a diamond had such a dark appearance, even though the pattern was considered a very good representation of a diamond with “optical symmetry,” many observers did not place it in the top category. This has been accounted for in our system.
[ top ]
How does “optical symmetry” relate to the symmetry that GIA currently reports, one aspect of finish?
To understand this issue, in fact three different meanings of symmetry need to be explored here: mathematical (or modeling) symmetry, grading (or physical) symmetry, and “optical symmetry.” These different meanings have different implications.
- A computer modeled, or virtual, diamond has “perfect” mathematical symmetry – all facets of a given type are evenly spaced around the outline and have exactly the same angle to the horizontal. All facets are perfectly flat, are shaped correctly, and relate perfectly to each other. The relationships of the facets in three dimensions (3D) are displayed well in most two-dimensional (2D) projections.
- The symmetry of a real diamond is graded for finish by examining the diamond’s surface, identifying major and minor symmetry faults (like table-off-center, misalignment between crown and pavilion, or a misshapen bezel), and assigning an evaluation. Multiple symmetry faults can interfere with each other such that some kinds of 3D mis-orientation (that is, 3D skew) are not easily seen from any single 2D view.
- Optical symmetryrequires a special viewing device that surrounds the diamond. The light passes through the diamond and interacts with the viewing device to create a pattern, either face-up or face-down, for the observer. The consistency and balance of this pattern results from the 3D symmetry of the diamond; the exact appearance of the pattern is related to the proportions of the diamond. Some kinds of 3D skew that are not easily observed on the surface of the diamond are revealed by the lack of optical symmetry. In addition, some corrections of 3D skew leave minor (finish grading) symmetry faults on the surface of the round brilliant.
In other words, there is not always a direct relationship between optical symmetry and graded (surface) symmetry, so that a diamond with high optical symmetry does not necessarily also have excellent graded (surface) symmetry, and vice versa.
[ top ]
What is the effect of the mounting and the background against which the diamond is viewed on the appearance of a round brilliant cut diamond?
Although we mostly studied the appearances of loose diamonds in our research, we did discover some interesting effects of mountings and backgrounds in the process. We concluded that while the trade usually uses a white tray or folded white business card to make judgments in appearance, certain judgments could be impaired by the use of the white background. For instance, some pattern features that are dark due to light leakage are less obvious when diamonds are observed on a white background. Conversely, observing diamonds on a black background over-emphasizes those appearance features. A neutral gray background returns an appearance similar to the appearance in many mountings.
[ top ]
In addition, the effect of hand oil, etc., on the appearance of a diamond can be drastically negative. For a diamond to look its best, it should be cleaned regularly. See “The Optics of a Dirty Diamond,” at http://www.gia.edu/research/1383/2273/article_detail.cfm
Last Revised on 03/21/05
You may contact us by email at DiamondCut@gia.edu
