The Artist “Obsolution” Project
Format
SOECS Senior Project Demonstration
Faculty Mentor Name
Michael Doherty
Faculty Mentor Department
Computer Science
Abstract/Artist Statement
Computers have traditionally been used as cognitive logical engines for solving difficult problems, but conventional wisdom says they lack the ability to be creative. This application is designed to prove that a computer can tackle creative problems by having an application generate digital art via artificial intelligence. In order to tackle this digital art generation problem, the application should be able to generate its own art based on guidelines. The application takes search terms from the user for what the final piece will look like, and uses these to find a goal image on the internet for what the final artwork will look like. The final artwork is a collage of smaller composing images, also found online by specified user search terms. The program utilizes a genetic algorithm to create a new collage of smaller images that will ultimately look like the goal image. Canvases are randomly generated in the beginning of the application and those that match the goal image closest on each iteration of the genetic algorithm are used to create new canvases. This process is repeated continuously until the canvases that the algorithm creates are similar enough to the goal image. This method allows our application to act in a manner that appears to be incredibly creative, breaking through the presumed barrier of creativity in computer science.
Location
School of Engineering & Computer Science
Start Date
7-5-2016 2:30 PM
End Date
7-5-2016 4:00 PM
The Artist “Obsolution” Project
School of Engineering & Computer Science
Computers have traditionally been used as cognitive logical engines for solving difficult problems, but conventional wisdom says they lack the ability to be creative. This application is designed to prove that a computer can tackle creative problems by having an application generate digital art via artificial intelligence. In order to tackle this digital art generation problem, the application should be able to generate its own art based on guidelines. The application takes search terms from the user for what the final piece will look like, and uses these to find a goal image on the internet for what the final artwork will look like. The final artwork is a collage of smaller composing images, also found online by specified user search terms. The program utilizes a genetic algorithm to create a new collage of smaller images that will ultimately look like the goal image. Canvases are randomly generated in the beginning of the application and those that match the goal image closest on each iteration of the genetic algorithm are used to create new canvases. This process is repeated continuously until the canvases that the algorithm creates are similar enough to the goal image. This method allows our application to act in a manner that appears to be incredibly creative, breaking through the presumed barrier of creativity in computer science.