- 最后登录
- 2017-9-18
- 注册时间
- 2011-1-12
- 阅读权限
- 90
- 积分
- 12276
- 纳金币
- 5568
- 精华
- 0
|
1 Introduction
In order to synthesize attractive facial expressions, it is necessary to consider detailed expression changes such as facial wrinkles. Nevertheless, most techniques of expression synthesis (i.e. blend shape, image morphing, simulation of mimic muscle and so on) focus entirely on large-scale deformation of a face and ignore small-scale details such as wrinkles and bulges. Therefore, hand crafted work of skilled artists is inevitable to make a face attractive finally, unless using a huge photographic equipment.
By using Expression Ratio Image [Liu et al. 2001], which shows wrinkles as illumination variation between two facial images of a same person with different expressions, it is possible to provide expression wrinkles to facial textures. Whereas, we have to collect suitable ratio images to create desired facial expressions beforehand. Dutreve et al. [2011] developed a method for automatically applying the optimum normal map obtained from a ratio image to a facial model with the animation of expression. The problem is, however, that the method is effective only against a common skeleton- based model.
In this paper, we propose a novel technique which automatically generates detail wrinkles on a face texture from only a few points movement of a face. By using our technique, personal characteristics of expression are cloned on any face model. As a result, facial animation is getting more impressive. seIn our method, the database of wrinkle maps, which is cons***cted by ratio images of several expressions, is used as a means to append small-scale details to a target facial texture. Synchronized with large-scale deformation of a face, an optimal wrinkle map is automatically generated by auto-regression with vertex movement and is mapped to the texture of an input face. |
|