中文摘要：

在布料动画领域,细节变形褶皱的快速逼真模拟是一个具有挑战性的问题.多精度网格布料动画技术,在不同变形区域采用不同密度的网格,由此保证布料运动的丰富细节和更高的动画合成效率.已有的研究工作,主要针对多精度网格布料动画的可计算问题,即在动画过程中动态获取顶点或面片曲率,依据手工设定的阈值进行动态迭代精化,而对于有效性问题,即如何预测布料运动趋势并据此合理修改网格精度,未深入研究.针对该问题,在对布料运动过程中弯曲变形特征进行分析的基础上,提出一种基于实例数据分析的多精度网格布料动画方法.首先,选取一段持续的高精度布料动画作为实例数据,分析并提取弯曲变形模式;其次,以弯曲变形模式作为启发信息,结合有效的精化策略以及精化原则,对相应低精度布料模型进行逐层精化,从而构建多精度网格布料几何模型;最后,建立多精度网格各层质点的受力关系,并通过增加边界约束和质量约束,构建可适用于多精度网格的布料动画模型.实验结果显示,文中预构建的多精度网格模型可用于同类相似运动驱动下的布料动画计算,能够在保持较多细节变形的前提下,有效提高计算效率.

英文摘要：

In the field of cloth animation, one of the challenging problems is simulating fine-details of deformation in efficiency. Multi-resolution mesh technique that mesh density depends on the deformation of each region can be applied in cloth animation so as to maintain rich details without reducing computing efficiency. Most of the existing work usually focused on how to dynamically refine mesh and how to compute numerical system. Few researchers consider how to construct the right multi-resolution mesh through observing and predicting the trend of deformation. In this paper, we propose a novel method to animate multi-resolution meshes based on cloth animation instances. First, a specific type of high-resolution cloth animation is chosen as an instance, from which deformation trend is analyzed and deformation modes are extracted. Second, the corresponding low-resolution model is refined based on the extracted deformation modes using right refinement strategy. Finally, the force is analyzed and the dynamic system is constructed for the multi-resolution model, in which the boundary constraint and mass constraint are considered. Experiments show that the constructed multi-resolution model is right for the similar human motions as instance, and not only are fine details maintained but also the computing efficiency is improved.