基于运动捕获数据的运动LOD技术研究

Research on Motion LOD Technique Based on Motion Capture Data

通过运动捕获技术获得的动画具有高度的逼真性,但由于其数据量大,限制了虚拟场景中动画角色的数量,因此虚拟场景中角色简化的主要目的是在不影响视觉效果的基础上降低角色的渲染花费。将几何LOD思想引入角色运动,对运动LOD进行分类,提出利用关节点的旋转信道计算帧差距来生成和控制不同层次的运动LOD模型,分析不同运动行为选择LOD模型的标准,离散LOD模型和连续LOD模型都得到了很好的简化效果,运动LOD简化技术是提高渲染速度、增加角色数量的有效方法。

The motion capture technology by which can obtain perfect reality animation is applied in computer human animation widely, but the number of animation characters is limited because of large data came with motion capture technology. So the goal of simplify of characters in virtual environment is to reduce the cost of rending the characters but not affect the vision representation. Introduced LOD to character animation, classified the motion LOD, presented frame distance which compute with the rotation channels of joints to generate and control different motion LOD models, and analyzed the criterions of LOD model choosing for different motion behavior, our methods achieved good effect both in disperse LOD and progress LOD, motion LOD is the effective method for improving the rending speed and increasing the number of characters.