Temporal Coherency in Painterly Rendered Computer Animation Using a Cognitive-Based Approach

This thesis proposes a solution to augment temporal coherency in painterly rendered computer animation sequences, using a computer engineering approach. Painterly is a cognitive knowledge-based parameterized Non Photo-realistic toolkit for creating artistically rendered still imagery. Therefore, it is incapable of maintaining temporal coherency of rendered animation frames. Consequently, movies rendered by Painterly demonstrate a significant amount of flickering. We proposed and developed CPA - a system to enhance temporal coherency in the sequences rendered by Painterly. CPA utilizes Painterly’s cognitive and perceptual knowledge space and induces coherency in the outputted results, by controlling and executing the main part of frame synthesis process. We created cognitive-based painterly rendered sequences which showed a good deal of improvement in maintaining temporal cohesiveness. Furthermore, by incorporating the element of ‘time’ in Painterly’s frame synthesis process, we expanded its scope from being a still-oriented and state-less toolkit to a more multipurpose and state-full system.