实时变形是计算机图形学研究的热点问题之一,复杂物体的实时变形至今仍未得到很好的解决.从物理变形方法和多分辨率网格编辑技术的优点出发,提出了一种适合于复杂弹性物体的实时变形算法.在预处理阶段,将原始精细网格模型进行简化以建立其基网格表示,基于基网格对模型的局部细节特征进行编码;在实时绘制阶段,在基网格上进行物理变形操作,并通过变形后的基网格和细节编码重构出变形后的精细网格.以上过程充分利用图形硬件的并行处理能力,利用像素处理器进行大部分计算操作.实验结果表明,该算法在变形过程中较好地保持了物体的局部特征,适合于表面细节复杂物体的实时变形应用.
Real-time deformation is one of the hot research issues in computer graphics and it is widely used in video games, virtual surgery and virtual reality. However, physically-based deformation can be computationally expensive, so the problem of real-time deformation for complex objects is yet to be solved. In this paper, a new method of real-time deformation is proposed, which integrates the advantages of physically-based deformation and multi-resolution mesh editing and works well with complex elastic objects. In the preprocessing stage, the algorithm simplifies the original detail mesh to generate a base mesh representation adaptively, and then encodes position and normal of the vertices of the detail mesh based on the base mesh. In the real-time rendering phase, the physically-based deformation is performed on the base mesh, and with the deformed base mesh and the encoding geometry details, the deformed detail mesh can be easily reconstructed. In order to improve the rendering speed, the rendering stage sufficiently utilizes the parallelism ability of graphics hardware and most calculation is executed by the fragment program. Experiments show that the method preserves the local details relatively well during the deformation procedure and it is suitable for the applications of real-time deformation of objects with complex surface details.