Particle Swarms and Aesthetic Virtual Photography

by William Barry (william.barry.mail AT gmail . com)
Supervisor: Brian Ross

This research presents an approach to automated photography in a virtual environment. A particle swarm optimization (PSO) algorithm is used in a virual 3D environment. Each PSO member is a camera, and the swarm is a collection of photographers. A number of constraints or objectives are given to the system, in order to search for a suitable image in the environment. Objectives include ones that encapsulate aesthetic composition rules, such as the rule of thirds, and horizon line. Colour matching and subject matter are also included. A new multi-objective PSO algorithm is presented, which is capable of finding solutions that satisfy multiple objectives (sometimes 10 or more).

Various images and animations from this research are available below.

Publications

"Generating Aesthetically Pleasing Images in a Virtual Environment Using Particle Swarm Optimization",
William Barry, MSc Thesis, Brock University, Dept of Computer Science, 2012. (PDF)
"Virtual Photography using Multi-objective Particle Swarm Optimization",
William Barry and Brian J. Ross, GECCO 2014, pp.285-292, Vancouver, BC, July 2014.
Presentation slides. (pptx)

Examples

Image gallery.
Animations:
- Sunrise (mov)): This is one of the particle's camera viewpoint as it moves in the swarm, looking for a good photograph.
- Sunrise swarm (mov): This shows the entire swarm searching for a good image.
- Space (mov)): Another particle's view.
- Complex environment (mov): Yet another particle's perspective.

Back up: http://www.cosc.brocku.ca/~bross/