Cinematography is the task of making movies, which includes camera work, scene transition, lighting, and coloring. This task can provide dramatic effects for text animation, or the automated generation of animation from a written story. Consequently, animation with dramatic effects if they conform to the intentions of the story writer can deliver what the story conveys in an unambiguous manner. This thesis presents a method that applies cinematography to text animation, focussing in particular on camera work and scene transition, by using natural language processing techniques to extract from a written story the elements relevant to cinematography and reflecting them onto the animation for the story.
The method extracts the elements which are necessary for camera work in two steps: 1) The method first assigns particular techniques of camera work to the semantics of natural language verbs, that indicate actions of characters, according to the pre-defined associations between techniques for camera work and semantic classes of verbs; and 2) the method adjusts the assigned techniques for camera work, with respect to the movements, positions, angles, and shots of camera, by analyzing the attributes of characters which play relevant roles in the actions indicated by the verbs. The method further extracts the elements relevant to scene transition by identifying the temporal relationships between the story-line events, where such relationships are critically represented in temporal adverbial phrases, conjunction items, and temporal aspects of verbs. The method utilizes a full-fledged natural language grammar formalism, or a combinatory categorial grammar, to extract such elements of cinematography from a written story.
Furthermore, the resulting system provides a user-friendly interface by way of which users are able to select the most relevant cinematography from candidates for the purpose of improving the interactivity between the system and the users, and of providing the diversity in animation generation. I also show the results of comparison between the animation generated by the system and those of other text animation systems from the perspective of how clearly the resulting animation delivers what the story conveys.
영상기법은 텍스트 애니메이션에 극적인 효과를 줄 수 있는 영상제작장치로서, 카메라 기법, 장면전환기법, 조명, 색채 등의 장치들을 포함하고 있다. 이야기글 속에서 작가의 의도를 정확하게 전달하기 위해서 나타나는 극적 장치들을 애니메이션 상에 효과적으로 표현하기 위하여 본 논문에서는 영상기법을 도입하고자 한다. 특히 이야기 글로부터 애니메이션에 필요한 정보를 추출하기 위해서 자연언어처리 방법을 사용하며, 이 정보들을 영상기법의 일부인 카메라 기법과 장면전환기법을 사용하여 표현하고자 하였다.
