Novel animation and rendering approaches to cartoon

Abstract

2D animation production has always been time-consuming and labor-intensive. This thesis investigates several processes in the animation production pipeline that are especially laborious. A 2D animation system has been created to provide more efficient method of performing those tasks. A tremendous amount of research work has been done in the creation of the system, including user-controllable automatic inbetween generation, automatic coloring, smart stroke input, high quality stroke rendering to provide accurate visual feedback and distributed framework to reduce computational time. Inbetweening is one of the most time-consuming processes of 2D animation production. Automatic inbetween generation addresses this issue, but it remains a tough issue in research. Most existing methods apply interpolation based on either pixel intensity of raster images or mathematical functions of vector representation, but neglect information in the drawing itself, e.g. shapes or motion of characters. That often results in distorted lines and shapes as well as unsmooth motion. This thesis contributes significant improvements to automatic inbetween generation algorithms. They include parameterized non-linear interpolation, feature point preservation and junction preservation. Coloring is another time-consuming process in 2D animation production. Great amount of research has been done intending to achieve automatic frame generation. This thesis proposes vector-based region recognition and region matching algorithms.