Generation of abstract geometric art based on exact aesthetics, gestalt theory and graphic design principles

In this thesis artificial intelligence ideas are applied to the domain of fine arts and especially modern art. First, we take a closer look at avant garde art movements of the late 19th and first half of the 20th century. After that, we make an analysis of the knowledge on which this art movement is partly based by considering the fields of aesthetics, gestalt psychology and graphic design. Having formalised general ideas about what a well-formed painting should consist of, we then look at ways of incorporate these ideas in a model for generating a composition. We design a formal framework to which we map the domain concepts. Based on the framework, we make a top-down knowledge decomposition. To demonstrate how our ideas can be applied in a practical situation, we have implemented a prototype system. Theoretically, this system is split up into the front-end part, in which the actual output is generated, and the back-end part, in which artificial intelligence techniques are applied to the actual concepts of composing an artwork. The front-end is partly based on the multimedia generation system called Cuypers, which was developed at the Centrum voor Wiskunde en Informatica (CWI) in Amsterdam. Cuypers was made to generate dynamic presentations and therefore generates XML, which subsequently is transformed into a desired format (XHTML, TIME, SMIL) using XSLT transformation stylesheets. Our system generates output in the Scalable Vector Graphics (SVG) format, which is an XML based standard for vector graphics and animation. The back-end part is based on the formal ideas about art described above. It is implemented in Eclipse Prolog. Finally, we discuss the artistic significance of our results. We conclude this thesis by discussing the advantages and disadvantages of the conceptual decisions, as well as suggesting directions for future research, including ideas for the evaluation of the generated compositions