Evolving stylized images using a user-interactive genetic algorithm
Steve Bergen · 2009
The application of evolutionary computation to art has produced very interesting results with respect to the generation of images [5]. Systems that rely on automatic image generation, where the computer is relied upon for the production of creative images, often fall short when results are presented to a user for feedback, as results in the field of visual arts are highly subjective. Many systems incorporate user-guided evolution to remedy this situation, requiring the user to supervise the evolutionary process at every step, though this task is largely repetitive. This paper introduces JNetic, a highly user-interactive image evolution system that does not rely on user feedback or supervision. It can be run in two modes, automatic and interactive, using color distance between a source image and generated image as a fitness function in automatic mode. JNetic is a tool for artists, created with user intervention and interaction in mind, but ultimately focusing on the automatic evolution of images.