Content uploaded by Tony Cristian Dumitriu
Author content
All content in this area was uploaded by Tony Cristian Dumitriu on Jan 30, 2018
Content may be subject to copyright.
© 2017 Ed. Univ. „Al. I. Cuza” Iași. All rights reserved
Corresponding author’s e-mail: dtonycristian@yahoo.com
Available online at http://geology.uaic.ro/auig/
Analele Stiintifice ale Universitatii “Al. I. Cuza” din Iasi
Seria Geologie 63 (1–2) (2017) 25–35
AUI
GEOLOGIE
3-D minerals. Auxiliary material for the Physical Geology classes
Dumitriu Tony–Cristian1, Balan Iulian–Vasile1
1 “Alexandru Ioan Cuza” University of Iaşi, Department of Geology, 20A Carol I Blv, 700505
Iaşi, Romania
Abstract
The paper presents a method for enhancing the learning process of mineral species by the
students of the first year in Geology, at the Physical Geology classes. However, the method can
be applied in many other fields that study physical objects. To be able to help students, the
photogrammetry techniques were used, together with a game engine platform, in order to create
a digital atlas containing 70 3D mineral samples. The atlas provides students with essential
information for those minerals, along with the opportunity to study some of the mineral’s optical
properties, interact with the sample and perform measurements.
Keywords: 3-D minerals, physical geology, first year students, photogrammetry, Unity 3D.
1. Introduction
Teaching geology is both fun and
satisfying as you expect to see your
students becoming experts in a field that
eventually will lead them to aid the
society through their activities. However,
sometimes, making the students under-
stand some parts of geology can be very
difficult, especially today when technol-
ogy can be a distraction. To be able to
counteract these distractions, we came to
the conclusion that it is necessary to use
the same “weapons”. Interestingly enough,
we later found out that students also
expected this to happen and they were
much happier to be able to learn by using
these methods.
The idea of the project was born in
early 2016, after discussions with the stu-
dents of the first year, regarding alterna-
tive methods for studying the minera-
logical samples in the “Physical Geology”
laboratory. Considering that this labora-
tory is also used by other students, the
time assigned to those of the first year is
rather short. Moreover, the collection of
samples that the students of the first year
can use has quite a limited number of
specimens for each mineral species.
26 Dumitriu T.-C. and Balan I.-V.
AUI–G, 63, 1–2, (2017) 25–35
Because of these factors, the first year
students always had some difficulties in
learning all the minerals, in order to be
able to recognize them later in their work.
It became clear that an additional solution
had to be found to resolve this problem, a
solution that will enable the study of
samples at home, without taking or
damaging the samples and having, at the
same time, all the information at hand.
The proposed solution was to create a
digital database with all the samples
needed to study, in a 3-D format and with
all the complementary information. This
digital database can be considered a digital
atlas which allows users to interact with
the samples in different and useful ways,
for better learning. The methods used for
achieving this goal was the 3-D modeling
technique known as “Structure from
Motion” Photogrammetry and the game
creation platform (Unity3D Game Engine).
Previous works of Minocha (2013)
and Minocha et al. (2014) have shown us
how using game engines for different
teaching purposes can improve the stu-
dents’ knowledge and help them to over-
come some other difficulties.
In regard to photogrammetry, there are
many papers which contribute to the field,
of which only a few refer to its
contribution in teaching, e.g., Kosmatin
Fras and Grigillo (2016) and Nikolić et al.
(2012). Very recently, an author has
begun a parallel work with us (https://blog.
sketchfab.com/photogrammetry-for-the-
classroom-3d-scanning-for-geology-and-
paleontology/), and has described a method
to obtain 3-D models from samples and
place them online, but without any
possib-ility to extract measurements or
any other information which can be
crucial in a mineralogical study.
2. Methods
In order to build up the “3D Mineral
Atlas”, the photogrammetry techniques,
with their related tools, as well as a game
engine platform known as Unity 3D were
used.
The entire process can be separated
into six major stages, which have their
own subdivisions: the testing stage, pho-
tography stage, photogrammetry stage,
cleaning stage, game engine stage and the
dissemination stage.
2.1 Testing stage
In the testing stage (Fig. 1), we have
looked for the best way to take photos
(sample position, light setup, camera posi-
tion) of the mineral, so that the final mod-
el would be as close to reality as possible,
with utmost errors as possible, yet keep-
ing the time needed to create it short. We
found out that a good balance between
time and quality (in both taking and pro-
cessing photos) is achieved with about
150 photos for each side of the sample,
taken from two different angles. The pro-
cessing of the 150 photos for each side of
the sample by using Photoscan and other
Fig. 1 Setup for testing different positions and an-
gles for photo taking.
3-D minerals. Auxiliary material for Physical Geology 27
AUI–G, 63, 1–2, (2017) 25–35
softwares like 123DCatch, RecapPhoto,
VisualSFM, resulted in two separate 3-D
models, which would have been later
aligned manually through a somewhat
lengthy process that often may return a
poor result. Trying to avoid as much as
possible the human interaction in the
modeling process of the samples, the best
method was to position the sample sup-
port in a white environment with diffuse,
white light (white led bulbs with white
paper wrapped around) and take two sets
of photos, i.e., at two different positions
(top tilted view and front slightly tilted
view) of the sample and at an interval of
about 5.6° (Fig. 2). This way, we were able
to model the whole sample in an automa-
tic way, using Photoscan, without the need
for further alignment of the two halves.
2.2 Photography stage
Using the method discussed previ-
ously, in the photography stage, four sets
of photos (of 64 photos per set) were
obtained, with a total of 256 photos for
each studied sample (Fig. 3). The next
step was to process the resulting photos
using Photoscape software, by batch
modifying each set to increase sharpness
and contrast. Adding sharpness and con-
trast to the photos is necessary in order to
help Photoscan in the photo aligning
stage. To have a close to reality texture
and color of the sample, the original
texture used in the texturing process of
the models was kept.
2.3 Photogrammetry stage
As shown above, in the photogram-
metry stage, a 30 day trial version of the
Photoscan software has been run on a
Graphic Station (Intel Core i7-5820k,
Nvidia Quadro K5200, 64 GB DDR4
RAM and a KINGSTON SSD). The soft-
ware reconstructs the 3-D model of the
Fig. 2 Final setup with diffuse white lights, a tripod
and a turning table for the sample.
Fig. 3 Examples of photos taken for the sodalite
sample. 12 photos from a total of 256 photos.
28 Dumitriu T.-C. and Balan I.-V.
AUI–G, 63, 1–2, (2017) 25–35
sample throughout four main steps: photo
alignment (which generates a sparse point
cloud), dense cloud generation, mesh
generation and texture generation. Photo-
scan also provides the possibility to use a
fifth step (chunk aligning), which is only
necessary for manual alignment of two
separate parts of the model, while only
one cannot treat the model as a whole. As
previously discussed, the need for the
fifth step was eliminated by choosing
good conditions for photography, in order
to eliminate the human factor from the 3-
D reconstruction of the sample.
For the first step, the photo alignment,
the “Accuracy” was set to “Medium”, “Pair
preselection” to “Generic”, “Key point
limit” to “0” and “Tie point limit” to “0”.
If the photos are taken and prepared cor-
rectly, the photo alignment step should
pass without any intervention and it should
generate quite a detailed sparse point
cloud (Fig. 4). The process for each sam-
ple photo alignment took about 40 min-
utes (more or less, depending on the detail
density of the sample). The number of
points in the sparse cloud is different from
one sample to another. For instance, a
sulphur sample (named “SULPHUR 1” in
the application menu) has 263,457 points,
a fluorite sample (named “FLUORITE 2”)
has 112,901 points and a magnetite sam-
ple (named “MAGNETITE 2”) has 243,003
points.
Fig. 4 Sparse point cloud generated for the soda-
lite sample (294,818 points).
Fig. 5 Dense point cloud generated for the sodalite
sample (5,357,726 points).
Fig. 6 Mesh generated for the sodalite sample
(2,359,212 vertices and 4,589,169 faces). Fig. 7 Mesh with textures generated for the sodalite
sample.
3-D minerals. Auxiliary material for Physical Geology 29
AUI–G, 63, 1–2, (2017) 25–35
The second step in the 3-D recon-
struction was the dense cloud generation
(Fig. 5). This step was the most time-
consuming, as it took an average of about
3 hours to complete with “Quality” setting
set to “Medium” and “Depth filtering” set
to “Mild”. The resulting number of points
from this process (dense point cloud) for
the same samples discussed above is
3,964,489 for sulphur, 658,325 for fluorite
and 6,268,345 for magnetite.
In the third step, the mesh generation
process (Fig. 6), the “Surface type” was
set to “Arbitrary” and “Face count” to “0”.
The process lasted for an average of 2
minutes and it generated 3D models with
various numbers of faces, depending on
the number of points obtained in the
previous step. For the sulphur sample we
obtained a model with 2,273,695 vertices
and 4,449,556 faces, for fluorite a model
with 1,347,746 vertices and 2,621,776
faces and for magnetite a model with
2,529,779.000 vertices and 4,907,332.000
faces. It can be presumed that the higher
the number of points from the point cloud
and implicitly the higher the number of
faces from the 3-D reconstructed mesh,
the better is the quality of the 3-D model
of the sample (Tab. 1).
The last step, the texture generation
(Fig. 7), took on average of 15 minutes to
process; it began with the replacement of
the photos used for 3-D reconstruction
with the original ones, which have not
gone through contrast or sharpness en-
hancement. In this stage, the original pho-
tos for each sample was used to keep the
original texture and color of the sample.
Photo replacement was carried out by
selecting all the photos from the “Photo”
window inside Photoscan and changing
their path with right mouse click “Change
Path…” command. The setting used for
texture generation process were: “Generic”
for “Mapping mode”, “Mosaic” for
“Blending mode”, “16,384 × 16,384” for
“Texture size” and “Yes”, which enabled
both “Color correction” and “Enable hole
filling”.
2.4 Cleaning stage
From the previous stage, 3-D textured
models were obtained for each sample
captured earlier by camera (Fig. 8).
Although the 3-D models had a high
resolution, the 3-D reconstruction process
was not perfect and it generated either
artifacts or holes in the created models;
the imperfections were generated in most
part by the lack of information in some
photos, due to the reflectance of the sam-
ple’s surface. The “errors” were corrected
by post-processing, with help of Autodesk’s
Remake software (formerly known as
Memento), through manually filling of
Fig. 8 Examples of 3-D models obtained for the
samples. a – textured 3-D model of the magnetite
2 sample; b – only the 3-D mesh of the magnetite
2 sample; c – textured 3-D model of the sulphur 1
sample; d – only the 3-D mesh of the sulphur 1
sample; e – textured 3-D model of the fluorite 2
sample; f – only the 3-D mesh of the fluorite 2
sample.
30 Dumitriu T.-C. and Balan I.-V.
AUI–G, 63, 1–2, (2017) 25–35
holes or erasing the useless parts (Fig. 9).
Using the same software, we corrected the
orientation of the model, rotated it in the
wanted direction and scaled it according
to the direct measurement of the sample.
Afterwards, the cleaned 3-D model meshes
were resized to about 100 mb/model, in
order to be properly imported into the
Unity3D Engine, still keeping the original
size of the texture.
2.5 Game engine stage
The now cleaned, scaled and resized
3-D models were imported into the
Unity3D Engine, where the application
that will be disseminated among students
was developed. We choose to create the
application with five windows, having
different menus and buttons, which help
choosing what to visualize. From the be-
ginning, we wanted to create the menus in
both English and Romanian language
(Fig. 10) so that the students can learn the
information in their native language and
understand the terms in an international
language as well. The next window shows
a menu with every modeled sample
arranged in groups (Fig. 11) that follow
the classification proposed by Gaines et
al. (1997). From this menu, a mineral
from one group can be chosen, so that a
different window is opened, allowing the
student to decide either to interact with
the 3-D model, or read more information
and see different pictures of that mineral
(Fig. 12). In the 3-D model window, the
student can use the mouse to Rotate, Pan
or Zoom, in order to settle the sample on
any side; he also can use a keyboard key
and mouse to measure the distance be-
tween any two points on the surface of the
sample (Fig. 13). The final menu was the
information menu, where students can
read synthesized information about that
mineral and visualize photos from other
mineralogical collections (Fig. 14). Infor-
mation for completing this menu was pro-
cessed using Gaines et al. (1997) and
Ianovici et al. (1979) publications, along
with https://www.mindat.org website.
Regarding the photos from other
collections they were taken from
https://www.mindat.org and used in com-
pliance with the license terms. Also, every
mineral information menu and photo has
its own link to the website information
and photo pages, respectively.
2.6 Dissemination stage
Because the application developed
through the present study uses high reso-
lution 3-D models and a consistent vol-
ume of information and pictures, its final
size is about 7 Gb. Therefore, we could
not develop a version that could be placed
and opened directly on a website. Since
Fig. 9 Manually filling the holes of a 3-D model
using Autodesk Remake functions.
3-D minerals. Auxiliary material for Physical Geology 31
AUI–G, 63, 1–2, (2017) 25–35
Fig. 10 Choosing the language from the start menu
in the Unity3D application.
Fig. 11 The mineral group menu.
we did not want to reduce its quality, we
built it up as to be used only directly on a
personal computer, by downloading the
archived application from a cloud,
unzipping and running it up on the user’s
computer.
To be able to keep a track of how
many users download and potentially use
this material, we also developed a simple
Google Form that after registration gener-
ates a password, which must be used in
unzipping the application.
The final process is that of accessing
the link to the Google Form, completing
the form and receiving a password and a
link to the application folder from where it
can be downloaded, unzipped and started.
3. Results and conclusions
In the end, we managed to develop an
application similar to a digital atlas with
70 3-D samples of minerals and varieties,
having information, pictures and website
links for each one. The students may down-
load and use the application by complet-
ing a Google Form (https://docs. google.
com/a/geology.uaic.ro/forms/d/e/1FAIpQ
LSdaz3DPgWOeJadrTNTDjxQbq6xSLO
X-dWETaEmYoZC02RlFNA/viewform),
which will provide them with both a
download link and a password to unzip it.
After distributing the application to
students, we almost instantly noticed an
improvement in their understanding of the
optical properties of the minerals, as they
were able to study the samples for a longer
time and in more detail, during individual
learning. Having constant and continuous
access to the collection, the students were
Fig. 12 An example of a mineral sample’s menu.
Realgar-Orpiment sample menu.
32 Dumitriu T.-C. and Balan I.-V.
AUI–G, 63, 1–2, (2017) 25–35
Fig. 13 Application menus. a – 3-D menu of realgar-orpiment sample; b – measurement
menu of realgar-orpiment sample.
Fig. 14 Application menus. a – information menu for orpiment mineral with corresponding
mindat.org links; b – photos menu for orpiment mineral with corresponding mindat.org links.
able to choose the perfect time for them to
study the samples, fact that would not
have been possible in the laboratory, due
to the limited amount of time allocated for
each class and limited number of samples
that can be studied at once. Moreover, the
application constantly provides the infor-
mation side by side with the 3-D model,
making it easier for the students to concen-
trate on the important parts of the class.
At the end of the semester, after the
practical exam, we also noticed a consid-
erable increase in their knowledge and in-
terest, which in turn led to better grades.
Further work on the application will
enable us to increase the quality of the
existing 3-D samples, add more samples
from our own collection and even in-
tegrate it on our department’s website.
Acknowledgements
The authors want to express their
gratitude towards Nvidia Co., for their
hardware support without whom the
achievement of this paper would not have
been possible.
3-D minerals. Auxiliary material for Physical Geology 33
AUI–G, 63, 1–2, (2017) 25–35
References
Gaines, R.V., Skinner, H.C., Foord, E.E., Mason, B.,
Rosenzweig, A., King, V.T., 1997. Dana's New
Mineralogy: The System of Mineralogy of
James Dwight Dana and Edward Salisbury
Dana, 8th Edition. John Wiley & Sons, 1872 p.
Ianovici, V., Ştiopol, V., Constantinescu, E., 1979.
Mineralogie. Editura Didactică și Pedagogică,
Bucureşti, 827 p.
Kosmatin Fras, M., Grigillo, D., 2016. Implemen-
tation of active teaching methods and emerging
topics in photogrammetry and remote sensing
subjects. Int. Arch. Photogramm. Remote Sens.
Spatial Inf. Sci., XLI, B6, 87–94; accessed at
https://doi.org/10.5194/isprs-archives-XLI-B6-
87-2016.
Minocha, S., 2013. 3D virtual geology field trips.
2nd Monthly International Workshop on Science
Exibits in online 3D environment, Abyss
Observatory in Second Life (3D virtual world).
Minocha, S., Davies, S.J., Richardson, B., Argles,
T., 2014. 3D virtual geology field trips: opportu-
nities and limitations. Computer and Learning
Research Group Conference. The Open Univer-
sity, Walton Hall, Milton Keynes, UK.
Nikolić, O., Pejić, P., Krasić, S., Nikolić, V., 2012.
Application of modern methods of photogram-
metry and software packages in teaching. In:
Proceedings of International Scientific Conference
People, Buildings and Environment 2012, 2,
799–804.
https://www.mindat.org
34 Dumitriu T.-C. and Balan I.-V.
AUI–G, 63, 1–2, (2017) 25–35
Tab. 1 Information about the number of vertices and faces, surface areas and volume measurements for
each 3-D reconstructed sample
Name Vertices Faces Surface area (cm2) Volume (cm3)
Graphite 1,193,841 2,305,183 15,245 112,258.315
Sulphur 1 2,273,695 4,449,556 390 459.800
Sulphur 2 2,387,714 4,656,170 640 813.790
Orpiment-realgar 3,434,426 6,676,778 738 1,232.522
Sphalerite 1 2,014,410 3,925,996 262 205.749
Sphalerite 2 2,955,228 5,756,720 428 586.503
Galena 1 941,521 1,801,816 154 93.877
Galena 2 2,145,103 4,199,240 133 105.626
Molybdenite 1 2,456,976 4,798,532 302 341.834
Molybdenite 2 1,125,835 2,188,904 271 298.003
Pyrite 1 2,247,560 4,409,847 18,947 200,712
Pyrite 2 797,709 1,538,304 80 46.811
Pyrite 3 644,536 1,181,620 77 23.840
Chalcopyrite 1,324,649 2,573,902 215 212.058
Realgar 3,434,426 6,676,778 738 1,232.522
Stibinite 1 2,644,876 5,033,968 20 2.070
Stibinite 2 1,779,902 3,462,388 208 132.741
Hematite 1 1,720,759 3,356,706 495 641.458
Hematite 2 2,059,736 3,995,467 312 345.381
Magnetite 1 1,981,759 3,867,554 290 235.368
Magnetite 2 2,529,779 4,907,332 258 284.418
Fluorite 1 1,531,464 2,967,954 152 122.885
Fluorite 2 1,347,746 2,621,776 187 166.573
Halite 1 2,191,493 4,268,448 381 451.665
Halite 2 2,548,004 4,991,868 692 1,387.598
Sylvite 566,787 1,097,120 36 14.171
Aragonite 1 1,744,868 3,395,052 206 177.868
Aragonite 2 2,093,751 4,050,732 411 447.928
Aragonite 3 2,120,848 4,135,783 242 259.247
Azurite 773,697 1,500,822 72 39.474
Calcite 1 1,619,668 3,157,998 195 185.744
Calcite 2 2,658,688 5,182,405 302 319.472
Malachite 2,024,263 3,941,708 198 181.994
Rhodochrosite 1,700,855 3,324,624 297 329.260
Siderite 1,129,800 2,177,599 64 21.764
3-D minerals. Auxiliary material for Physical Geology 35
AUI–G, 63, 1–2, (2017) 25–35
Name Vertices Faces Surface area (cm2) Volume (cm3)
Witherite 2,186,170 4,263,004 248 262.761
Baryte 1 219,395 421,340 26 9.341
Baryte 2 2,403,067 4,664,912 346 426.306
Gypsum 1 3,502,321 6,813,206 306 319.936
Gypsum 2 1,816,616 3,508,660 366 308.931
Gypsum 3 1,280,104 2,482,660 167 129.140
Actinolite 1 1,725,665 3,359,141 213 172.408
Actinolite 2 624,737 1,204,774 40 13.658
Albite 2,554,791 4,980,940 168 131.688
Asbestos 2,121,175 4,126,350 327 260.443
Biotite 405,179 786,028 44 16.711
Kaolinite 1,405,930 2,740,486 187 176.523
Chlorite 902,003 1,722,714 133 66.490
Quartz (chalcedony) 2,369,836 4,613,108 429 661.897
Quartz 1 1,885,613 3,665,916 163 118.353
Quartz 2 1,497,958 2,904,574 185 142.735
Quartz 3 803,184 1,559,872 43 18.377
Quartz 4 212,268 407,417 24 10.205
Quartz 5 76,844 141,918 4 0.297
Epidote 1 2,198,583 4,278,150 230 203.346
Epidote 2 2,556,496 4,991,080 280 335.750
Garnet 1 2,505,449 4,886,042 560 691.100
Garnet 2 1,642,789 3,209,444 204 199.582
Hornblende 2,171,637 4,222,030 517 497.610
Muscovite 1 1,885,500 3,642,018 255 189.446
Muscovite 2 1,601,670 3,116,334 210 170.752
Nepheline 1,936,276 3,780,066 322 364.300
Olivine 1,787,696 3,483,996 443 654.023
Opal 1 2,266,104 4,374,066 156 106.207
Opal 2 415,051 803,944 27 8.050
Orthoclase 1,592,716 3,084,287 61 29.456
Titanite-tourmaline 1,046,244 2,024,983 114 68.102
Sodalite 2,359,212 4,589,169 411 435.093
Talc 738,805 1,427,511 66 24.367
Tremolite 1 3,061,502 5,940,878 339 350.270
Tremolite 2 1,283,817 2,465,992 233 163.212
