Developing of Diamond Cut Grading System by MSU (OctoNus and GC MSU) Computer Tools

Thanks to Garry Holloway, FGAA, DipDT, Australia
for help during preparing this material

 

Introduction

There are two tasks in diamond cut study:

  1. First, or direct, task. Creating a system of appraising the cut quality.
  2. Second, or reverse, task. Creating an instrument or a methodology to grade a cut of real stones according to the above system by the cheapest and the most reliable way. The second task cannot be solved before the first one.

Some organizations are trying to solve the second task first so the market uses their appraising system and their instrument. It is not the way we want to follow.

Different methods of solving the reverse task presented at the market:

 
Human-guided
Automatic
Analysis of diamond image taken with a structured lighting
FireScope
BrillianceScope
Measuring diamond proportions
Proportionscope
Sarin

The main part of this paper is devoted to analysis of possible solutions of the first task.

 

Evolution stages of knowledge on diamond cuts - solution of the first, or direct, task

Author
Year
Method
Achievements
Drawbacks

Cutters

Before 1919

Empirical

Pattern created, proportions found

High price of diamonds makes impossible a complete study by one company (even for a single pattern).

Tolkowsky

1919

Statistical review + ray tracing calculation
- in round,
- in marquise 

Tolkowsky - First page 

Tolkowsky - Comparison of the Theoretically Best Values with those used in Practice 

Selected one of all sets of parameters

Other sets of parameters were not found. Scintillation was not taken into account. The methodology of statistical researches is not described.

Other calculations (Harding, Dodson, Eppler, Eulitz, Vasiljev, Varshavsky etc.)

ХХ century

Ray tracing calculation by different methods 

Algorithms, software, instruments created for researches on this subject

Researches did not leave the theoretical stage. Their work had little impact on the market.

Gemologists + traders

middle-end of ХХ century

Classification of components of diamond perception 

Two triads of notions are formed:
One based on features; Proportions, Symmetry, Polish
Second based on benefits; Scintillation, Brilliancy, Fire

Technicians find the first triad measurable. There is no methodology to evaluate the second triad.

Devices for appraising (Firescope, Brilliancescope etc)

From 1984

Real diamonds are analyzed with specific fixed lighting

Firescope 

Quantifiable Symmetry Analysis 

Repeatable results on real stones

Lighting conditions in the devices differ from those in real world. Factors of different nature (proportions, polish, absorption, inclusions) are mixed together.

Device for measuring cut proportions (Sarin)

1991

Scanning and creating a model of a real diamond

Main parameters are measured quickly and precisely enough

Only measures parameters.

AGS

1996

System of appraising cut of a round brilliant cut

Method created for appraising diamond beauty by measuring main parameters of the cut (features)

Assumption that separate each feature affects diamond beauty independently.

GIA

1998

Light return coefficient is determined on the base of computer 3D model of a diamond and illumination 

Light return graphs for ranges of main cut parameters

Incorrect illumination used. Only light return is analyzed. Selected light return model not related to percieved reality.

Garry Holloway

1984-2000

Empirical methods

Results of investigations

In 1984 – 1990 he observed (using a Firescope™) a link between crown and pavilion angles.

In 1992 he estimated a 2:1 relationship between crown:pavilion angles either side of Tolkowsky’s ideal crown and pavilion angles. In May 1999 he publicised this 2:1 concept. In January 2000 he adjusted this to 4:1 after communications with MSU scientists.

The serious progress in empirical findings of beautiful diamonds have been achieved from Tolkowsky ages. The correlation for beautiful stones not precisely describes areas of diamond parameters with equivalent appearance.

MSU

1999

Computer model: illumination - diamond - observation 

Analytical instruments for appraising cut quality

Different viewing conditions are taken into account. Modeling and calculations in real-time. Possibility to scan real diamonds and to perform calculations on their 3D models.

Preliminary results promising but experimental testing of good and bad zones is needed.

 

Factors of human perception of diamond appearance

Factors of human perception of diamond appearance that determine cut quality can be divided into three main groups:

Table legend:
Green color denotes features already implemented in Bril software Yellow marks directions in which there is some progress Red is for features that are not yet implemented

 

1. Parameters of light sources that characterize the illumination:
1.1. Source type (its spectrum, luminance, and size)
1.2. Position of light source
1.3. The number of light sources

 

2. Optical effects exhibited by the diamond itself:
2.1. Light return - the capability of the diamond to return a fraction of the incident light to the observer's eye
2.2. Fire - the capability of the diamond to disperse a white light into iridescent colors in a direction able to be perceived by the observer
2.3. Color - the result of the dependence of the light absorption factor of the gem upon the wavelength of the incident light
2.4. The pattern of facets and highlights within the diamond and their symmetry
2.5. "Dynamic Contrast" - the rate of change of the diamond appearance while stone is rotated (either light source or the observer is moving)
2.6. Clarity - the prefered absence of inclusions that absorb or scatter light
2.7. Surface lustre, flatness of facets, quality of polish, and surface smoothness

 

3. Viewing conditions:
3.1. Position of the observer relatively to the diamond
3.2. Psycho-physiological particularities of the observer :
3.2.1. Stereoscopy of vision
3.2.2. Pupil motion To other page
3.2.3. Eye sensitivity
3.2.4. Eye adaptation
3.2.5. Color perception, effect of the background
3.2.6. Eye resolution
3.2.7. Spatial and temporal contrast
3.3. Psychological criteria of diamond beauty perception.
3.4. The existing stereotypes, fashion on the market.

 

 

Classification of components of diamond perception

Triad: illumination, diamond, eye.

To create a grading system with repeatable results it is necessary to fix (standardize) illumination and observation conditions, distance to the diamond, mono or stereo mode of viewing, dynamics of stone movement relatively to the observer and of illumination. Also, relative input of scintillation, brilliancy and fire to diamond's beauty should be evaluated.

 

MSU methodology of cost reduction for creating cut quality grading system

 



1. Selecting "bad" zones in automatic mode for the whole range of parameters using light return analysis, fisheye, "kozibe", etc. (a method was created for round cut diamonds and can be accommodated for fancy shapes)

- Light Return Picture To other page (illustration from "Diamond Cut Study in MSU"),

- "Kozibe" + Fisheye + Light Return Picture To other page

Fire and scintillation calculations (Round 1 Iteractive model, Round 2 Iteractive model To see illustration, Marquise Iteractive model) for the whole range of cut parameters. Selecting zones of equivalent cut quality for given parameters. Merging such zones with those of valid parameters that were determined on the first stage.

2. Selecting borderline and definitive diamonds (control stones for cutting).

3. Cutting of experimental stones

4. Scanning of resulting diamonds, creating their 3D models To other page.

5. Visual testing of control diamonds. Testing how a human evaluates their photorealistic images and synthesized movies.

6. Grading of control stones. Correction if necessary for Fire estimation formulas To other page and new formulas for Scintillation should be developed (Bril).

 

Some features of OctoNus Ltd. software products

BRIL

DIAMCALC

3D-BOOK

PACOR

Software for cut study inside a company

Commercial software for modeling diamonds with built-in systems of grading cuts

Interactive learning software on diamonds

Hardware and software package for marking large rough diamonds

Creating realistic images of diamonds, ray tracing, stereo mode, calculation of diamond weight

 

Calculating parameters of a cut quality by automatic search

Creating video movies to demonstrate scintillation.

A course on diamond grading

One of features is scanning and building a model of real diamonds

Calculating a “photorealistic” (realistic photo-like) image of a diamond for different illumination conditions, shapes and absorption spectra.

Loading and editing different shapes created by GemCad software

Interactive tutorials

In addition to polished diamonds, it works with rough and partially polished diamonds

Demonstration of distribution of light leaving the diamond (can be used for quantitative evaluation of scintillation)

Possibility to work with 3D models of real diamonds that are created by PaCor software

Photo galleries

Possibility to determine inclusions’ positions

 

Example: Understdanding the Bowtie effect in diamonds

There is no real diamond for which Bowtie effect is present due to leakage of light through the pavilion, as all Bowtie effects (there are only two main types of them) appear due to partial shielding of a light source by a human head. We know two types of Bowtie effect:

1. First type occurs when a pavilion angle is about 45°. This case is not very interesting, as it rarely happens in practice

2. Second type appears when a pavilion angle is 39-40°, a crown angle depends on it (for example, when a pavilion angle is 40°, then a crown angle should be in 29-36° range. If a pavilion angle is 39°, a crown angle should be 36-42°. The ranges of a crown angle are specified for a photo camera with 14° angular size). The second type is more common in practice, as it is closer to round diamond proportions.

Conclusions for Bowtie effect of the second type:

1. If a pavilion angle is fixed (less than 43°), one can avoid Bowtie effect by selecting a proper crown angle. It is important both for cut grading and for cut prognosis

2. A notion that the Bowtie effect depends only on a pavilion angle is false. It can be seen especially clearly for a 39-40° pavilion angle

3. When the observer's head is moving further from the diamond, the Bowtie effect can diminish up to complete disappearance. If we think that the look of a diamond from a larger distance is more important than from a close range, the presence and value of the Bowtie effect should not affect our estimation of a diamond (for the second type of the effect).

The reasoning above illustrates possibilities of MSU technologies with application to appraising fancy shapes on example of the well-known Bowtie effect.


The authors are:
Sergey Sivovolenko, OctoNus Software, model and calculations
Yurii Shelementiev, Gemology Center of MSU, gemology