teatching_geometry_in_design_school
teatching_geometry_in_design_school
To cite this Article: Feijs, L. & Bartneck, C. (2009). Teaching geometric principles to design students. Digital
Culture & Education, 1:2, 104-115.
URL: http://www.digitalcultureandeducation.com/cms/wp-content/uploads/2010/01/dce1015_bartneck_2009.pdf
Abstract
Keywords:
Design, math, tessellation, Escher, geometry
Introduction
One of design’s main goals is to give form to products and communication. However,
most design students approach form intuitively, neglecting the understanding of the
underlying geometry. Adobe Illustrator, for example, gives powerful tools to students to
design geometrical shapes, such as Bézier curves (see Figure 1).
The mathematical principles of Bézier curves and other geometrical functions remain
hidden. A quadratic Bézier curve, for example, is the path traced by the function B(t),
given the points P0, P1, and P2:
Teaching such abstract and technical topics to design students is challenging and we
experimented with new approaches for industrial design students (Vlist et al., 2008). In
104
Teaching geometrical principles to design students
this study, we describe a new method of teaching geometry to industrial design students,
but we believe that it can be used for other design disciplines as well, such as graphic
design or fashion design.
Geometry is of course a topic characteristic of most high school maths
education and therefore we focus on slightly more advanced topics than, for example,
the Pythagorean theorem. When it comes to geometrical principles of design, the golden
ratio is frequently brought forward. It is surprising that a book entitled “Geometry of
Design” (Elam, 2001) is limited to the Golden Ratio, root squares and the human
proportion. Moreover, the authors approach to directly apply these classic
proportioning systems to design classics indicates how much design is concerned with
the concretization and how little it is concerned with abstraction (Bartneck &
Rauterberg, 2007). We capitalized on this move towards the concrete by focusing on a
field of design in which geometry is the design: tessellation.
A tessellation is a collection of plane figures that fills the plane with no overlaps
and no gaps. Tessellations are used in industrial design, textile design and interior
design. Industrial design applies tessellations, amongst others, to tiles for bathrooms and
kitchens and to paving stones. The principle of tessellation has also found great
attention in manufacturing, since it minimizes the amount of wasted material when
punching pieces from sheet metal (Bigalke & Wippermann, 1994). These application
areas focus on the outer shape of the tiles. Sometimes tiles of different color are
combined to introduce more variations.
Textile design and interior design focus even more on the inside of the tiles.
They add decorative patterns onto the tiles, leaving the principle of the tessellation in
the background. Throughout the centuries, fabrics and wallpapers utilized tessellation to
create patterns. Maurice Cornelis Escher (1898-1972) took tessellation to the next level
by adding meaning to the tiles. The graphical elements on the tiles are no longer just
decoration, but they describe, for example, animals (see the fish in Figure 2).
Schattschneider (1997) explored the different combinations in depth.
Before describing the process of the course, we will first give a short introduction to the
different types of geometry, including group theory, matrices, and topology.
105
Feijs & Bartneck
Geometry
• Euclidian geometry,
• analytical geometry,
• transformational geometry.
If we are to assign one mathematician as the main or most famous person behind each
of these three views, then we have to name Euclid, Descartes and Klein. To simplify
matters we call them the inventors of Euclidian geometry, analytical geometry and
transformational geometry, respectively. Euclid (Greek: Εὐκλείδης) lived around
300BC and his main work is known as “The Elements” (Euclid, Heath, & Densmore,
2002). René Descartes (1596-1650) published his main filosophical work “Discours de la
méthode” (Descartes & Clarke, 1999) and the mathematical “La Géométrie” (Descartes et
al., 1659) both in 1637. Felix Klein (1849-1925) presented his influential “Erlangen
Program” in 1872 (Sharpe, 1997).
Euclidian geometry studies points, lines, circles, angles, lengths, areas, etc. It is
the heart of the geometry taught at secondary schools with nowadays aspects of the
other views blended in. Euclidian geometry can be studied on a purely axiomatic basis,
and although the mathematics community honors Euclid for this, modern primary and
secondary school teaching of geometry relies heavily on links to practical applications
and examples from optics, physics and everyday live. Particularly in the Netherlands,
this is the case (Feijs, 2005). Figure 3 shows one of the oldest fragments of a textbook on
geometry and indeed the text is about lines, segments, rectangles, sections etc. 1
The ideal media for this type of geometry are books and classrooms using pencil and
paper, with tools such as straightedge (ruler) and compass. It also goes very well with
traditional blackboard and chalk or modern white board and marker. Although things
1
translation: If a straight line be cut into equal and unequal segments, the rectangle contained by the
unequal segments of the whole together with the square on the straight line between the points of
section is equal to the square on the half”, [5]
106
Teaching geometrical principles to design students
can be done using computers, usually it means that someone has to program most of it
first, using some form of analytical geometry. A typical example is The Geometry
Applet by Joyce (1996). Plane geometry fits well with today’s flat computer screens and
ink jet printers. Important results include Pythagoras’ law, functions such as sine, cosine,
tangent and their inverses. These results are indispensable when designing tessellations.
Now we turn to the second view, analytical geometry. Descartes connected
geometry and algebra by coding the points of geometry as number pairs. Each point is
given an x and a y coordinate. In Figure 4 this approach is shown in action.
y x2+y2=4
2
-2 -1 1 2 x
-1
-2
The circle with radius 2 around the origin is described by an equation x2 + y2 = 4. The
circle can be shifted, for example, the substituting x → x–2 and y → y–2 and a bit of
algebra give the equation of the circle around {2,2}, which is x2 + y2 – 4x – 4y + 4 = 0.
Straight lines, ellipses, parabolas etc. each have their own type of equation. Geometry is
replaced by algebra. Deep geometric results can be obtained by unleashing the power of
algebra, for example to prove that one cannot construct a square whose area is equal to
that of a given circle using straight ruler and compass needs Galois theory (Galois,
1830a, 1830b), which is algebraic in nature. The ideal medium for analytical geometry is
anything that helps to solve equations. First pen and paper, later log tables, calculators
and now of course computers. When using computers, numerical procedures are useful,
symbolic procedures are useful and their combination is best.
Finally we address the third view on geometry, transformational geometry. As we
shall see, this is about symmetries, including the symmetries encountered in art, from
ancient times to Escher and beyond. It also is useful for modern physics, for example
quantum physics, but that is outside the scope of our teaching. In order to explain the
basic idea, we refer to Figure 5 (snapshot of Microsoft PowerPoint).
Consider geometric transformations such as rotation, translation and reflection,
for example those shown in the menu of Figure 5. The two-arrowed block shape can be
rotated left over 90° or right over 90° but in both cases it becomes a different shape.
When flipped horizontally or vertically it turns into the same shape again. The same
applies rotation over 180° (twice 90°). Therefore we can say that the two-arrowed block
shape is a symmetric shape. More precisely, each of the transformations which leave the
figure the same, is a symmetry of this figure. Formulated somewhat more abstractly, the
underlying mathematics is called group theory. Two operations can be composed, doing
one and then the other, which is a transformation again. For example rotate left over
107
Feijs & Bartneck
90° and then (again) left over 90° has the combined effect of the rotate 180°
transformation . 2
In two dimensions one can create patterns filling the entire plane such that the pattern
remains the same under certain transformations. For a specific pattern, the set of
transformations forms a so-called group. Next to translations, reflections and
translations one needs an extra transformation called glide-reflection. Group theory is in
the heart of studying regular patterns. There are three possibilities:
• If there are no translations, the patterns are so-called rosettes. There exist two
sets of rosette groups, either with or without reflections. They have applications
in art and architecture, most notably the rose windows in Gothic cathedrals. A
number of rosettes of logos, each with different symmetries, are in Figure 6. The
logos are Opel (180° rotation), Crosiers (180° rotation and reflections),
Mitsubishi (120° rotation and reflection), Yamaha (idem), and NATO (90°
rotation, no reflection).
• If there are only translations in one direction, the patterns are so-called friezes.
There exist seven distinct frieze groups. They have numerous applications in art
and architecture such as friezes of ancient temples and court buildings.
• If there are translations in two or more directions, the patterns are so-called
wall-paper patterns. There exist seventeen distinct wallpaper groups. They
appear on wallpaper indeed, in brickwork, in floor tessellations, in Escher’s work
and so on.
Figure 6: Five rosette logos with different symmetries (Opel, Canons Regular of the Order of the Holy Cross,
Mitsubishi, Yamaha, NATO)
There exist also non-periodic tessellations such as invented by Penrose (1974), which
have only been noted a few decades ago (folk wisdom of crystallography assumed such
patterns would not be possible). In our teaching we used the rosette and the frieze
2 A collection of such transformations is said to form a group if the following properties hold: Closure: if
two transformations are in the group, then so is their composition; Associativity: composing T1 with “T2
composed with T3” is the same as “T1 composed with T2” composed with T3; Identity: there is a special
transformation called the identity (doing nothing) which can be composed with any T with the effect of
T; Inverses: for any transformation T in the group there is a transformation denoted as T–1 such that T
composed with T–1 equals the identity and also T–1 composed with T equals the identity transformation.
108
Teaching geometrical principles to design students
patterns as easy forerunners of the more complex wallpaper patterns. We did so both in
the lectures and in the assignments given to the students. As literature we provided
Chapter I of the book Regular Figures by L. Fejes Toth (1964). The pedagogic idea
behind this choice is that the book is mathematically quite rigorous3 and that it is a good
exercise for the students to develop their skill of getting information from such a book
directly, instead of low-threshold and colorful Internet resources such as the Wikipedia
Wallpaper group article [ http://en.wikipedia.org/wiki/Wallpaper_group ]. Another
reason is that the idea of one transformation being transformed by another turns out
useful later, when implementing transformations. It is very practical being able to read
formulas like S2–1S1S2 and interpret them as action sequences to be programmed.
What is the ideal medium for studying and using this type of geometry? Again,
pen with paper and computer algebra are useful, but both the development of intuition
and the practical applications call for ways to make the transformations happen in
physical reality and before one’s eyes, or even better, in one’s hands. Shifting pieces of
paper one over another, transparent sheets, sliding and rotating images and semi-
transparent images in picture editors, physical tiles, paper and scissors are all very useful.
For certain Heesch tilings (see next sections for a detailed discussion of the Heesch
tilings) there exist a trick called “The Envelop Method” based on slicing 3D objects
(envelopes) into tile-able ground forms (Feijs, 2008). It is important to realize that the
student can not “see” the transformations in the same way he or she can see the points
and lines of Euclidean geometry.
3 To give an impression what we mean with “quite rigorous” we give a short quotation from Fejes Toth
(pages 11 and 12), the paragraph where it is explained why a rosette group can only have rotations around
a single centre: Such a group contains only rotations about a single centre. For, if there were two rotations S1, and S2 with
distinct centres O1 and O2, the transformation S1–1S2–1S1S2 of the group would be a degenerate rotation, by the additivity
theorem for angles of rotation. It cannot be the identity, since it displaces O1, into the image of O1, under S2–1S1S2, i.e. under
S1 transformed by S2. But the centre of this transformed rotation is the image of O1 under S2 i.e. a point different from O1
and therefore S2–1 S1S2 effects a change in O1 . Hence S1–1S2–1S1S2 would be a non-degenerate translation, contrary to our
assumption.
109
Feijs & Bartneck
deploys “mapping”, using the Mathematica operator /@. This operator is well known to
programmers in special languages such as LISP, ML or Clean but no such thing exists in
C, C++ or Java.
Figure 7: Matrices for identity, flipping, 90° rotation and 45° rotation.
This is a network of tile edges. This network has 32 vertices, 41 edges and 11 faces
(including the outer space). Now there exists a mathematical result about such networks
named after Leonhard Euler (1707-1783). It belongs to topology, that is the geometry of
properties which are preserved by “rubber” transformations. Euler found that V – E +
F = 2, where V is the number of vertices, E the number of edges and F the number of
faces. Indeed, 32 – 41 + 11 = 2. Heesch (1963) characterizes this network as type
333333, meaning that the six consecutive vertices when going round one tile, each
connect 3 edges. The above network has eight tiles, so a first very rough guess could be
that is has 10 x n vertices but of course most vertices are then counted three times
instead of once. And the first guess of 10 x n edges has to be corrected because edges
are counted twice. Going from 10 tiles to N tiles and inserting the corrections we find
that N.(⅓ + ⅓ + ⅓ + ⅓ + ⅓ + ⅓) – N.½.n + N = 2. For very large N, taking the limit,
we find ⅓ + ⅓ + ⅓ + ⅓ + ⅓ + ⅓ = ½.n – 1. This is for the given honeybee network,
but the same reasoning can be applied to find all networks of triangles (n = 3). Let the
three vertices have i, j and k edges. Now 1/i + 1/j + 1/k = ½.3 – 1. This is only
possible for certain numbers of i, j and k. For example ¼ + ⅛ + ⅛ = ½, so i = 4, j = k
= 8 is a possibility. Thus the network type 488 is found. Working further in this way
110
Teaching geometrical principles to design students
only 11 networks are possible. Finally Heesch (1963) integrates this theory with
wallpaper theory to get 28 tiling types. He describes them in a prescriptive style (in
German). This is what we gave to the students (actually we gave access to all of
Heesch’s book, but we have no indication that they read anything except the
prescriptions). An example is Figure 9. The network type is 43433. The tile itself is given
a type code CC4C4C4C4 where C refers to 180° rotational symmetry and C4 to 90°
symmetry. These prescriptions are very practical to use. They could be used without any
group theory or network theory, but we did present the essentials of those theories
nevertheless.
C
A
B
90°
E C4 C4
90°
D
C
Net:43433, 4 Orientations B
90°
C4
C
E
C4
D
Corner of a net
Figure 9: One of the 28 Heesch tile descriptions (Heesch & Kienzle, 1963), translated from German by the authors.
111
Feijs & Bartneck
Workflow
a b
c d
Figure 10: Workflow of tessellation. (design by Wouter Widdershoven)
The students prepare their tessellation in Mathematica. The lines they draw define
the paths that the laser cutter will use for the cutting and engraving. The colours of
the lines define if a certain line is used for cutting or engraving. For the engraving,
the line thickness defines the thickness of the engraving. For the cutting, an ideal
value of 0.001 mm is used. It is important that the students do not simply create a tile
and repeat it, since this would create overlapping lines. The laser cut would cut this
line multiple times, which takes far too much time and can reduce the overall quality
of the tiles. The students export the Mathematica graphic as an encapsulated post
script (EPS) file and final adjustments in Illustrator. The illustrator file is then
imported into CorelDraw and transferred to the Laser Cutter control software
through a virtual printer driver. Figure 10 shows the workflow that the students go
through. First, they create a tile (a) which is then repeated using rotations and
translations in Mathematica (b) before the necessary colours and thicknesses are
assigned (c). The graphic in (c) is then sent to the laser cutter. Last, the students
create a coloured version of (d), which makes it easier to puzzle the pieces together
(see Figure 11).
The tiles are then mounted to foam board or another plate of Perspex.
Transparent Perspex is more difficult to mount, since the mounting plate and the glue
shines through. Additional examples of the students’ work are available at
http://www.bartneck.de/2008/05/21/the-golden-ratio-course/.
112
Teaching geometrical principles to design students
Figure 11: Perspex tiles cut with a laser cutter. (Design by Wouter Widdershoven)
Conclusions
We described a new approach for bringing the mathematical foundations of geometry
into the focus of design work. The approach is new to the best of our knowledge. One
possible reason is that the tools used are relatively new and moreover the community of
mathematicians and the community of industrial designers are usually very separated.
The students did not use common 2D design software packages, such as Adobe
Illustrator, to create a tessellation, but we asked them to define the tessellation in the
abstract language of math. This truly brought mathematical foundations of geometry
into the focus of the course and Mathematica is one of the most accessible tools for
design students to express their ideas in math. Of course the students could have
stopped their effort after the production of a digital version of their tessellation, but
going the extra mile of cutting real tiles with a laser cutter had a tremendous effect on
the motivation of the students. The final tessellations have been hung at the walls of our
department and it represents a great reward to our students. The Perspex tiles moved the
abstract geometrical principles out of the computer and into the real world. The
tessellations are physical objects to which the students can relate. They are not just an
abstract ideas, but products that can be touched and experienced. The students
responded that they appreciated the brushing-up of their math, but most of all it was
fun. They truly enjoyed understanding the mathematical principles of vector graphics, as
it is used by Adobe Illustrator and other design software. Moreover, it empowered them
to create their own tessellations more easily. The patterns used in the currently
fashionable handbags (e.g. Louis Vuitton) are now trivial to them and they can create
advance tessellations for the design of textiles, bathroom tiles and visual design in
general. Our approach facilitates design because of four reasons. First it empowers the
design students because they become familiar with Mathematica and related tools, which
allows them to address other modeling problems as well. Secondly it allows the students
to design patterns whose aesthetic qualities can be added or embedded in other works
of design. Thirdly it gives them additional knowledge about patterns as they appear in
nature and in technical domains. Fourth it invites them to learn more about Escher,
Japanese art and similar cultural topics relevant for design students.
We can also conclude that the technology for creating tessellations has
developed rapidly. The math tools and graphic design tools allow the creation of
tessellations within minutes and the remaining limiting factor is the creativity of the
designer. The development of affordable laser cutters also has a tremendous effect on
tessellation: the cutting line is negligible. It is no longer necessary to calculate in the
113
Feijs & Bartneck
width of the cutting line and true tessellations can be cut in one run. In the future we
intend to extend the module to include aperiodic patterns, such as Penrose tilling. We are
also further exploring the usage of semi-transparent Perspex that can then be glued
against windows, creating an effect similar to windows in cathedrals.
References
114
Teaching geometrical principles to design students
Biographical statement
Dr. Christoph Bartneck is an assistant professor in the Department of Industrial Design at
the Eindhoven University of Technology. He has a background in Industrial Design and
Human-Computer Interaction, and his projects and studies have been published in
various journals, newspapers, and conferences. His interests lie in the fields of social
robotics, Design Science, and Multimedia Applications. He has worked for several
companies including the Technology Centre of Hannover (Germany), LEGO
(Denmark), Eagle River Interactive (USA), Philips Research (Netherlands), and ATR
(Japan). Christoph is an associate editor of the International Journal of Social Robotics.
Email: [email protected]
Prof. dr. ir. Loe Feijs (1954) studied Electrical Engineering at TU/e where he graduated in
1979 in the group Information and Communication Theory. He joined Philips
Telecommunications Industry, later AT&T-Philips Telecom and in 1984 Feijs changed
to the Philips Natuurkundig Laboratorium In 1990 he obtained a Ph.D. in Computer
Science of TU/e. In 1994 Feijs was appointed part-time professor at TU/e Mathematics
and Computer Science. From 1998 to 2001 he was scientific director of the Eindhoven
Embedded Systems Institute and in 2001 he was appointed full professor for the chair
Industrial Design of Embedded Systems.
Email: [email protected]
115