高级检索
马千里, 刘涛, 王攀, 刘瑜, 李思昆. 基于时空一致性的非结构化网格时变流场高效体绘制方法[J]. 计算机辅助设计与图形学学报, 2011, 23(11): 1816-1824.
引用本文: 马千里, 刘涛, 王攀, 刘瑜, 李思昆. 基于时空一致性的非结构化网格时变流场高效体绘制方法[J]. 计算机辅助设计与图形学学报, 2011, 23(11): 1816-1824.
Ma Qianli, Liu Tao, Wang Pan, Liu Yu, Li Sikun. High-Efficiency Volume Rendering of Unstructured-Grid Time-Varying Flows Using Temporal and Spatial Coherence[J]. Journal of Computer-Aided Design & Computer Graphics, 2011, 23(11): 1816-1824.
Citation: Ma Qianli, Liu Tao, Wang Pan, Liu Yu, Li Sikun. High-Efficiency Volume Rendering of Unstructured-Grid Time-Varying Flows Using Temporal and Spatial Coherence[J]. Journal of Computer-Aided Design & Computer Graphics, 2011, 23(11): 1816-1824.

基于时空一致性的非结构化网格时变流场高效体绘制方法

High-Efficiency Volume Rendering of Unstructured-Grid Time-Varying Flows Using Temporal and Spatial Coherence

  • 摘要: 时空一致性是时变流场的重要性质,也是加速时变数据可视化算法的关键.以硬件加速的光线投射算法(HRC)为框架,设计并实现了一种基于时空一致性的非结构化网格时变流场高效体绘制方法.首先提出一种分析非结构化网格单元和顶点数据时间一致性的方法,分别建立单元和顶点数据时间表,以降低绘制过程中的计算开销;然后设计一种单元和顶点数据相分离的GPU纹理结构,并采用一种小巧的单元梯度矩阵来降低显存开销;同时,设计了一种合理的数据调度策略,既能有效地避免绘制停顿,又使显存纹理结构更为紧致、高效.实验结果表明,该方法不仅明显地提高了绘制效率,而且具有更优显存空间利用率,能实现更大网格规模的非结构化网格时变流场数据体绘制.

     

    Abstract: The temporal and spatial coherence which is an important characteristic of unsteady flows plays an essential role in visualizing time-varying fields.This paper presents an approach for high-efficiency volume rendering of unstructured time-varying flows using the temporal and spatial coherence on the framework of the hardware-based ray casting algorithm(HRC).Firstly,a method is provided to analyze the temporal coherence of both the cell and the vertex data on unstructured grids.Then the cell and the vertex data temporal tables are built to achieve a lower time cost during rendering.Secondly,a novel texture structure is designed to separate the vertex data from the cell data,and a smart gradient matrix is used to reduce the pressure of GPU memory.The scheme of data management can effectively avoid rendering stalls and lead to a compact and efficient storage.The experiments demonstrate that our approach not only gains a much higher efficiency than the existing method,but also achieves a lower space cost,allowing rendering time-varying data on a larger mesh scale.

     

/

返回文章
返回