EVA 05 Full Paper Submission


See Complete Results HERE

See also an example of evolving two original painter programs HERE



----------------------------------------- ------------------------------------------------------


Submitted Full Paper in COLOR can be downloaded here (MS WORD)





Paper Info:                             Steve DiPaola – Full Paper Submission Eva05 London           Page 1 of 2

Author:             Steve DiPaola

School & University:     School of Interactive Arts and Technology, Simon Fraser University

Address:                       10153 King George Highway, Surrey, British Columbia, V3T 2W1, CANADA

Email:                           steve@dipaola.org                 URL:                              http://www.dipaola.org


Full Paper Title:

Evolving Creative Portrait Painter Programs using Darwinian Techniques with an Automatic Fitness Function


Extended Abstract:

In evolutionary art, most systems evolve paintings by allowing the artist or viewer to selectively breed the computer artwork by hand from the current evolved population of paintings. Our evolutionary art system differs from this typical process in two significant ways: 1) it evolves painters not paintings and 2) it uses a ‘creative fitness function’ so no user intervention is needed to determine the most aesthetic choice, attempting instead to have the computer be creative on its own. Both of these approaches are significant research areas. To achieve these goals, a new type of Genetic Programming (GP) system is used called Cartesian GP, which uses typical GP Darwinian evolutionary techniques (crossover, mutation, and survival), but has several features that allow the GP system to favor creative solutions over optimized solutions including accommodating for genetic drift where different genotypes map to the same phenotype, visual mapping modules and a knowledge of a painterly color space.  Portrait painting was chosen for this project as it limits the space, weighs towards resemblance, and has a known sitter/painter relationship well suited for computer creativity. This work with its specific goal of evolving portrait painter programs to create a portrait ’sparked’ by the famous portrait of Darwin (the resemblance fitness function), speaks to the evolutionary processes as well as creativity, as seen by the early results where the evolving programs use recurring, emergent and merged creative strategies to become good abstract portraitists. Any of subsequent painter programs can be combined to reproduce new offspring.


Sampling of Results:         See additional and color samples here:  http://www.dipaola.org/evolve/eva05
The 'most fit' (resemble the Darwin portrait below better than their neighbors) of a population of these portrait programs are 'married' together to create 'more successful' offspring. These selected portraits below are in order, starting with the first population and moving in chronological order. I have selected a few of the 1000s, both those best at resemblance of the peers as well as "strange uncles" that were very off track from the dominant “resemblance” strategy but still compelling from a abstract portrait perspective. See website for more.

We start evolving portraits based on this picture (1), after a few populations, the color then curves emerge.
100s later, a first strategy appears: bands can resemble the strong vertical lighting of portrait, then they twist & curve.
Soon the bands, twists & curves strategies create the dominant image (1) below and the first "head shapes" appear.
As the early evolution progress slows, genetic drift brings in a colorful phase (more on the website) .      DiPaola_EVA
Then after days, this new ramped dominant strategy takes over (1 below) heralding in the soft blobby age. Page 2 of 2
The n ext major strategy addition is the left "raccoon patch" eye area and the right eye.
Still evolving, a more painterly age begins, combining the head shape with painterly surface (see site for more).