![](bg2.png)
ics decisions in a similar fashion as humans can”
(Hoenig, 2005). In the field of computational aesthet-
ics, evolutionary systems can play an important role,
by enabling the evolution of aesthetically pleasing
or innovative structures (DiPaola and Gabora, 2009).
Evolutionary art is characterized by the use of evolu-
tionary principles and natural selection as a genera-
tive process. One of the earliest applications of evo-
lutionary systems to generate art is the proposal of
Sims to use an EA to create complex images (Sims,
1991) or virtual creatures (Sims, 1994). In evolution-
ary art systems, the evaluation of the aesthetics can
be done using human feedback, with some interactive
evaluation of the population, such as (Ashlock, 2006;
Draves, 2006; Moroni et al., 2000) and (Sims, 1991).
It also can be achieved by using an automatic evalu-
ation of fitness, as presented in (Aguilar and Lipson,
2008; Den Heijer and Eiben, 2010; DiPaola and Gab-
ora, 2009; Li et al., 2012), and (Sims, 1994).
One of the main challenges in Evolutionary Art is
how to measure aesthetic value of a piece of evolu-
tive art. The source of this difficulty lies in the in-
herent complexity, subjectivity and dynamism of aes-
thetics. Nevertheless, a wide number of metrics has
been presented. According to (Galanter, 2012), these
measures can be classified into several categories in
several pieces of research. The first category involves
the evaluation of the aesthetics of a piece of art by a
formula or principle (e.g., pythagorean proportions).
Other measures apply certain principles of design,
such as the rule of thirds or color theory (e.g., us-
ing complimentary colors in Pop Art (den Heijer and
Eiben, 2012)), neural networks or complexity mea-
sures.
This classification also provides a sub-
classification for evolutionary systems. First, it
identifies interactive evolutionary systems, where the
fitness of the individuals is determined by human
agents. Another category is performance based goals:
certain properties of the art piece are evaluated and
optimized based on performance measures (e.g.,
usable surface in a furniture design generator). Other
systems use an exemplar (i.e., real world example)
as a way to measure the fitness of the individuals.
Finally, some models use the idea that the complexity
is directly related to aesthetics and follow the path
firstly stablished in (Birkhoff, 2003). Given the
multidimensional nature of aesthetics judgement,
multi-objective EAs are a straightforward option to
deal with this multidimensionality. Other extensions
to EA, like coevolution or agent swarm behavior, can
be used in evolutionary art systems.
A brief classification of the aesthetic measures
found in the evolutionary art systems mentioned in
the previous paragraph is shown in Table 1.
Several methods for the representation of the art
in evolutionary art have been proposed. In symbolic
expression, the genotype is a tree of expressions and
the phenotype consists in the image produced by the
evaluation of the tree. Shape grammars can also be
used as a formal description of the image. Previously
existing images can be used as a basis for the evo-
lution process. Finally, other representations can be
based on mathematical models, like fractals or cellu-
lar automata.
3 PROCESSING AND
HISTOGRAMS
In this section we describe Processing
1
and the his-
tograms used. Processing (Reas and Fry, 2007) is
a framework formed by a simple programming lan-
guage and an integrated development environment
(IDE) primarily created for electronic and visual
artists, designers, musicians, etc. Processing offers
the following advantages:
• Processing was created for artists, rather than pro-
grammers. So, it allows very complex draw-
ings and interactive applications with few lines of
code.
• It is an Open Source software (licensed under the
GNU Lesser General Public License), and counts
with a large development community.
• It is based on OpenGL, thus providing 3D accel-
eration.
• It also includes more than 100 libraries for video,
sound, physics, computer vision, networking, etc.
• Easy integration with Java, HTML5 and Android.
However, being a light framework, there exist
some disadvantages:
• More complex applications require more pro-
gramming skills.
• The calculations of large computer images are a
bit inefficient (although expert programmers can
manage OpenGL at low level to fix this).
There exist a lot of interactive artistic projects
made with Processing; examples include art
generation, artificial life, interactive music
and other. A good selection can be seen in
http://processing.org/exhibition/.
The Color module can be used to analyze images
taking into account their histogram. The color his-
togram represents the frequency of occurrence of each
1
http://www.processing.org/
TestingtheDifferencesofusingRGBandHSVHistogramsduringEvolutioninEvolutionaryArt
169