Artistic Content Representation and Modelling based on Visual Style Features
Thesis DisciplineComputer Science
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
This thesis aims to understand visual style in the context of computer science, using traditionally intangible artistic properties to enhance existing content manipulation algorithms and develop new content creation methods. The developed algorithms can be used to apply extracted properties to other drawings automatically; transfer a selected style; categorise images based upon perceived style; build 3D models using style features from concept artwork; and other style-based actions that change our perception of an object without changing our ability to recognise it. The research in this thesis aims to provide the style manipulation abilities that are missing from modern digital art creation pipelines.