Computer Assisted Cartoon Character Pose Editing And Interpolation

The animated cartoon has a history of more than one hundred years since the first cartoon film "Humorous Phases of Funny Faces" was produced in 1906. Nowadays the animated cartoon has been coming into our daily life, and some bewitching animated cartoons, such as "Brothers Cucurbit", "Tom and Jerry and" and "The Mickey Mouse and Donald Duck Cartoon Collections", are very popular and bring the growing youth much happiness. The animated cartoon is a dynamic film, which is a rapid display of a sequence of 2D images where each image is called as a frame. For a 15 seconds length animated cartoon displayed at the the rate of 8 frames per second, it contains almost 7200 frames. Furthermore, in each frame there are usually several characters where each one is associated with a specific pose. In the traditional production of animated cartoon, all the cartoon characters and their poses are manually drawn by animators. Obviously, the production of even a short animated cartoon requires a vast work of the animators. Thus to produce a high quality and long animated cartoon is a time, human resource and money consuming work!In this thesis, we investigate techniques of computer assisted cartoon character pose editing and interpolation, in order to improve the efficiency of the animated cartoon production. One class of techniques are related with the cartoon character manipulation by which users can create desired character poses via a simple click-and-drag way. The other ones are related with the character key poses interpolation by which the smooth and natural inbetween poses can be generated automatically or semi-automatically. The thesis is organized as follows:In Chapter 1, we first briefly review the history of the animated cartoon and introduce its traditional production process, and then give an overview of the computer assisted cartoon production techniques. We also introduce some representative softwares and analyze their advantages and disadvantages. Finally, we indicate the significances and difficulties of the research on the computer assisted generation of the cartoon character poses. In Chapter 2, we give a survey of the related work on the computer assisted cartoon character pose editing and interpolation, i.e. deformation and morphing. The deformation methods include free form deformation, skeleton driven deformation, physically based methods and differential domain based methods, and we make careful analysis and comparison among them. 2D shape morphing methods are closely related with the cartoon character key poses interpolation. Besides of the introduction of 2D shape moprhing methods, we also point out their advantages and disadvantages carefully.In Chapter 3, we propose three methods for cartoon character manipulation in the context of shape deformation. The methods provide efficient and flexible tools for the computer assisted generation of cartoon character poses.·Firstly, we propose a stiffness tunable 2D polygonal shape manipulation method. Using the method, the user can manipulate the shape in the "click-and-drag" way. In this method, we design a geometric energy function to avoid the shape distortion during the manipulation such that we can obtain the physically plausible shape deformations. The method also provides users a simple and intuitive interface to adjust the local and global stiffnesses of the 2D shapes, such that we can mimic the deformation effects of the shapes with different materials or inhomogeneous material. In addition, by employing fast summation algorithm, the method is accelerated to achieve real time response to the user interaction, which makes it possible for user to edit the cartoon character poses interactively.·Then we introduce a topology-aware shape manipulation method to address the geometry and topology inconsistent problem that may be occurred in the previously proposed stiffness tunable manipulation method. The method retains the major advantages of the stiffness tunable manipulation method: stiffness tuning and real time editing, but avoids effectively the unnatural deformation results. In addition, the new method can be straightforwardly extended to manipulate other 2D objects such as curves and stick figures. ·At last, we design a "general skeleton" based shape manipulation method for the cartoon character that is represented by several open or close curves. These curves may be disjoined, connected or overlapped. During the manipulation, the cartoon character's deformation is driven by the "general skeleton". In the method we introduce an energy function, which consists of rigidity, coordination and position constraint components, to control the deformation of the "general skeleton", such that the details and the relative positions of the features on the cartoon character can be effectively preserved. Finally we can obtain visually pleasant poses for the character represented by the complex geometric structure in a simple "click-and-drag" way.In chapter 4, we introduce the cartoon character key poses interpolation methods. The methods are closely related with the 2D shape morphing techniques, and include solutions to the feature correspondence and path interpolation problems. The methods can facilitate the animated cartoon production potentially.·We present an efficient and robust algorithm which can establish the feature correspondence between two polygons automatically. Furthermore, using the algorithm we can associate the features between two given key poses that are represented by several open or close curves in a semi-automatic way.·Then we introduce a two-level hierarchical interpolation method to interpolate the the cartoon character key poses. The method is suitable for not only the key poses represented by a polygon but also the ones that are represented by several open or close curves. During the interpolation, this method can effectively preserve the details and the relative positions of the features on the key poses. Finally, the method can generate the natural and smooth inbetweening poses automatically even for the neighboring key poses that differ greatly in their orientations.In the last chapter, we conclude our work, point out the problems that need to be solved and indicate the future work.