Animated Non-Photorealistic Rendering - Multiple Styles -

Abstract

This thesis presents an algorithm and its application for the artistic rendering of recorded video data based on non-photorealistic applications. The proposed method does not only work on a variety of artistic rendering styles for static photography but it can also be applied for the creation of artistic videos. To date, many algorithms have been introduced in the field of non-photorealistic rendering. For example, in a work by Crystal Valente, different artistic styles were emulated on static photographs. This project extended the methods to an algorithm for obtaining animations as artistic videos. Cartoon-like and comic-like styles are the two artistic styles introduced in this thesis. The goal in this project is to create animations in these two artistic styles. For creating successfully an artistic video, three key challenges are addressed: temporal (colour) consistency, stylistic flexibility, and scalability. Within this project, the main focus in video processing is on temporal consistency between consecutive frames. This leads to an improvement in visual quality. Stylistic flexibility aims at allowing users to experiment with different artistic styles to meet aesthetic requirements or individual preferences. Scalability is to control the time performance of the involved processes. Our work on addressing these challenges started with collecting image data, i.e. videos and photographs. Then we designed a method for video-based nonphotorealistic rendering from those input image data, either in cartoon-like or in comics-like style. The thesis demonstrates the benefit of the designed video-based rendering framework by comparing its results with results obtained from existing Android apps. Keywords: Artistic rendering, non-photorealistic rendering, animations, cartoonlike style, comic-like style.