Crystallography and Extrapolative Modeling

[Wilfred L. Guerin]

Crystallography and Extrapolative Modeling...
1: Overview of issue
2: Target material qualities *****
3: Extrapolation methods other than typical

We are tasked now with generating accurate 3d structural models of
natural crystals, specifically Asterated and Chatoyant optical factors
of natural minerals like Star Sapphire, Star Diopside, Cats and
Tiger's Eye, Synthetic and Natural Star Quartz (rutile/rutilated), and
my hope is that someone has worked with this topic especially in a
distributed and common manner.

The target systems are nothing more than a common digital camera
(possibly scanner of film) through a magnifying glass or microscope or
none at all.

Though scanning microscopes are fine for the task, the target audience
is the common jeweler and gemologist, moreso a neighbourhood rock
shop, and the materials are typically low cost and collector items.

We have a nicely centralized distributed processing system to manage
the data from imaging, however the sociological problems of the
jewelery industry typically prevent photographic images of rocks from
being published due to informal organized regulations in the market.

Obviously, if a 3d structure model can be pulled off at client side,
without significant external logistics (someone else's code, our
software, etc) being introduced to the machine that is likely also
doing billing and account management, it would satisfy all
requirements of research and offload the structure data and
characteristics without inducing mob concerns.

Although our code which typically handles reverse modeling for
FEA/engineering systems extrapolation is capable of handling this, the
distinction between normally-opaque and normally-transparent materials
warrants a separate technique optimized for the process.

*** 2:
The materials of interest, are those that generate natural spatial
characteristics that are typically known as the asterism (star on
sapphire) and chatoyance (tiger and cat's eye) which has depth
perceived from the reflection and refraction off various internal
crystal structures. Typically, these are small but humanly visible
structures of Rutile (SILK) which has a natural yellow appearance in
long strands forming partitions between the quartz/corundum crystal,
and reflecting the light back out.

Though the optical characteristic generates a very vivid 3d
"holographic" to the human eye, very few cameras, especially digital,
pick up on the fine particulates (the resulting blend at any
resolution is the same as sand at a beach) and thus it is troublesome
to model the structure of this at any resolution.

A common microscope can easily be used with automation and a common
digital camera to generate frame data sufficient to run normal models,
however, I am not in the position to distribute china market
microscopes with hacked automatic stage and scope with a quickiecam
tied on.

The lense characteristics can be modeled out, color/etc adjusted, and
the optical propagation of light through the structure can be reverse
modeled using the resulting shadow/hotspot in images...

The interest, obviously, is how to deal with both the camera pixle
resolution versus the optical characteristics, simplified imaging (we
use the targeting spheres stage for 3d, it works quite well), and
extrapolation of optical derivatives that would be used to model the
structure.

*** 3:
I have made a number of simple automated scanning systems, typically
used in the classroom with very simple protocols to reverse engineer
an old clock or other simple mechanical system, however the reverse
modeling of optics (especially lense) is computationally intensive,
especially when targeting fine particulates under the resolution of
the camera...

The intent, is to devise a system to extract all critical data for
this process from an extremely simple method, like pointing a camera
at a stage/rotary that has some balls on it, a known light source, and
getting various images.

<b>
More importantly, is the elimination of the affinity or representation
of the object itself and concentrate on an extrapolated structure that
represents the object sufficient to determine the crystal
characteristics of said object without awareness of the object itself
when not available...

Specifically, in all of these materials, their atomic and crystal
structure is quite simple, typically visible by eye or simple
magnification, the structure of the crystal is known, the critical
component is the amount of structural variation from an absolute
standard and a good estimate of the size of the smaller crystals and
their density.

The purpose is quite simple, as you see in Star Sapphire or Black Star
(india) Diopside, there is a 3d perceptual structure that is suspended
or projects illusionary from the material. Our intent is to accurately
model this system for public modeling of crystal structures and to
devise open rendering methods that are accurate.
</b>

*** 4:
Simply put, in the same manner as you would normally derive 3d
structure of lab specimins in slides, or anyone would generate 3d
extrapolated models of an environment, this project requires a
utilization of potentially poor imaging sources with a very humanistic
approach to acquire enough data to model at least the structure of the
object (normal 3d) and a good estimate of the optical qualities of the
internal structures.

From your experience in low resolution or optically skewed data,
especially depth stack slides or other transparent/translucent
materials, please hint at plausible methods to extract sufficient data
on target structures without handling the processing with the
excessively advanced systems and techniques im used to ;)

Thanks in advance for your input, and hopefully the resulting
optimized extrapolation method for crystalline and transparent/lens
materials will be of use to everyone...

-Wilfred L. Guerin
[hidden email]