Information Technology Reference
In-Depth Information
Painting Fool. The software will derive aesthetic measures via machine learning
techniques applied to the results of this crowd-sourcing activity. However, we will
attempt to avoid so-called “creativity by committee” by enabling The Painting Fool
to concentrate on those pictures which are liked and disliked by the crowd in equal
measures. In this way, its first learned aesthetic will hopefully be able to tell whether
a piece it produces is divisive or not, which is a start. We plan to enable the software
to invent and employ various aesthetic measures in a similar fashion.
It's our intention for the art produced by The Painting Fool to cause audiences to
engage their mental faculties, and not just to think about the fact that the pieces were
computer generated (although we advocate full disclosure of how the software pro-
duces its artwork, as described below). This will be achieved through the production
of intrinsically interesting work, which includes emotional content, interesting jux-
tapositions, social commentary, and so on. A level of audience engagement will also
be made through the software framing its pieces in various art-historical and cultural
contexts, and providing titles and wall-text to this extent. There are already art gen-
erating programs which achieve a good level of engagement with audiences, some
of which are described in other chapters of this topic. Indeed, there are many dif-
ferent kinds of conversations one can have with generative art pieces. For instance,
in some of the pieces produced by the
NEvAr
evolutionary art system described in
(Machado and Cardoso
2002
), rather than being driven by the user, the software uses
built-in fitness functions to search for art generating programs. When viewing these
pieces, one might be tempted to try and determine what the fitness function was
and how it is expressed in the pieces, in much the same way that one might try and
work out the aesthetic considerations going through a human painter's mind when
they painted their works. Concentrating on evolutionary art, other projects have ap-
pealed to the (un)natural world to evoke feelings in audiences. In particular, often
the fact that generated creatures (by for instance, Sims
1994
) and flora and fauna
(by for instance, McCormack
2008
) look so similar, yet dissimilar to real examples
of the natural world can lead to feelings of other-worldliness. McCormack's work
on evolutionary decay takes this further, via appeal to the art-historical mainstay of
mortality. Similarly, the software in the Mutator project as originally described by
Todd and Latham (
1992
) produces organic forms which can be unnerving—possibly
because of a similar effect to the well-known uncanny valley effect in video games,
where automated non-player characters get too close to being human-realistic, caus-
ing an uncanny, uneasy feeling in many people.
All of these approaches produce works which are thought-provoking indepen-
dently of their evolutionary genesis, and there are numerous other generative art
projects which produce interesting and culturally relevant artworks, with Romero
and Machado (
2007
) providing a good starting point for further reading. However,
authors such as Galanter (
2010
) point out that often the most interesting aspect of
evolutionary artworks is the process which went into producing them. This follows
a long line of art movements where the principle innovation has been the production
process (e.g. impressionism: painting
en plein air
to catch fleeting light conditions;
pointillism: painting with complimentary dots of paint, as per colour theories, to
produce more vivid pieces, etc.). We certainly advocate providing a description of
