Transcript Evolutionary Art

Evolutionary Art
(What we did on our holidays)
David Broadhurst
Dan Costelloe
Lynne Jones
Pantelis Nasikas
Joanne Walker
Introduction
• Project aims
– Develop some “nice” art
– Use Genetic Algorithms to evolve art
– Analyse the human-computer interface
Our Tools
• Graphical User Interface written in java
• Genetic Algorithm Engine
• Evolvable .gif images
Objectives
• Primary objective: “nice” art
• Secondary objective: modify GA engine
Art Requirements
• Needs to be simple yet attractive
• Evolvable – through parameters
• Written in simple java for easy
integration
Art Used
Kaleidoscope applet
– Simple shapes & bright colours
– Use of reflections and symmetry add
interest
– Animation gives extra dimension
– Small simple java applet
(Demo)
Chromosome Design
• Existing code used integer array
• Our parameters:
– Shape type (line / rectangle / circle / mix)
– Colour of shapes (10 x 255 colour palette)
– Background colour
– Symmetry style (horizontal / vertical /
diagonal / mix)
Experimentation
• Considered mutation rate and crossover
type
• Attempted to evolve a population of
solutions without circles
• Recorded speed of convergence
Mutation
• Guassian mutation
• Probability varied between 0 and 1
• As rate increased more generations
were required
• Convergence of other parameters
increased as mutation rate decreased
Crossover
• Considered three types:
– Single point
– Random N point
– Uniform
• Similar results obtained for each
Demo
Human Computer Interface
• Current layout is easy to use
• Can be time consuming after a few
generations
• Tournament style selection may be an
improvement
Conclusions
• Future work
– Implement other evolutionary algorithms
– Addition of visual effects on the animations
– Revision of Human Machine Interface
Thanks to….
•
•
•
•
Ben
Gusz
Bart (dude)
Andrew