Animation par l'exemple de texture vectorielle

Easy access to multiple screens for people and the growth in computational power in the last few decades promoted the emergence of more and more complex animated contents.This phenomenon creates a need for intuitive and efficient animation tools. In this thesis, we focus on methods allowing synthesis of animated textures, i.e. algorithms guided, or computer assisted technic to create animated textures. Such textures are widely used in numerous applications : wall projections on buildings, presentation enhancements, videogames, movies, etc. Although many research aims to synthesize animated textures, they focus mainly on creation and animation at the same time. Few are addressing the animation of already existing textures. We propose a system to address the problem of animating static textures by focusing on several key aspects of the textures and the animation process. Based on properties of the statistical distributions, the proposed approach allows intuitive animation of all elements in an input texture from a small example set of animated elements. The user has to specify the spatial structure of the example animation to form what we call the motion groups to guide a greedy search for similar dispositions based on spatial and color similarities. The animation of the static elements in the input texture can't be achieved by simply applying the same motion as a given example because of small variations in the spatial arrangement allowed by the statistical distributions of textures. An analysis of the user specified motion hierarchy and structure allows us to adapt the motion to the differences in spatial arrangements. Several constraints are established in the methodology in order to synthetize animations with a minimum rigid deformation of example motions. The proposed method shows promising results in both the similarity search and the adaptation of motions. However the result is dependent on the specification of the motion groups with enough differences from one another.