以直观、自然的方式来检索运动数据可有效地提高角色动画生产效率.提出一种通过现场示范表演来检索运动数据的方法,使动画制作人员能够直观、快速地从大容量运动数据库中找到所期望的运动片段.首先通过运动传感器获取用户的示范表演,然后利用经典的DTW时序数据匹配算法来实现基于样例的运动数据检索.通过一组观察实验对它的可用性进行分析和评估的结果表明,文中方法能够显著地提升运动数据检索的效率及用户体验.
With the increased amount of motion capture data, motion reuse plays a significant role in animation production, and motion retrieval in an intuitive and natural way is always preferred by animators. In this paper, we present a novel approach to retrieving motion with live performance. In the retrieval process, motion sensors are employed to capture the preformed movements as the query motion, and the matched motion clips are chosen by the classical DTW algorithm. A series of experiments are designed and conducted to evaluate its usability. The experimental results show that the retrieval scenario may allow the animators to intuitively locate the desired motion from the motion database and improves the user experience for content-based motion retrieval.