A parallel preintegration volume rendering algorithm based on adaptive sampling

Abstract

A parallel preintegration volume rendering algorithm based on adaptive sampling is proposed in this paper to visualize large-scale scientific data effectively on distributed-memory parallel computers. The algorithm sets sampling points adaptively by detecting the extremal points of a data field along the rays, so it can grasp the data variation exactly. After the data field is sampled distributedly on CPU cores, the resulting sampling points are sorted by piecewise packing orderly sampling points and then composited along each ray using the preintegration technique. In the algorithm, a static load balancing scheme based on information entropy is also proposed to balance the loads of both data reading and ray sampling. In addition, a mixed logarithmic quantization scheme is suggested to quantize depth distance so as to shorten the preintegration table while preserving the rendering quality. It is demonstrated that the presented algorithm can show inner features in a data field clearly and achieve a rendering speedup ratio between 1.8 and 4.4, compared with the traditional parallel volume rendering algorithm.

An objective-adaptive refinement criterion based on modified ridge extraction method for finite-time Lyapunov exponent (FTLE) calculation

“…Adaptive mesh refinement (AMR) is a method for adapting the accuracy of a solution within certain sensitive or turbulent regions of simulation [29,15]. This technique has been widely applied in CFD and proved to reduce the cost of calculation [37,25,18,21,1]. AMR has also been used in FTLE computation, which can enhance the efficiency of calculation by a large amount.…”

An Approach to Optimization of Ray-Tracing in Volume Visualization Based on Properties of Volume Elements

“…For example, in surface rendering such optimization schemes are widely used in different optimization approaches [8][9][10], such are level-of-detail optimization or culling-orientated optimization. Volume visualization has different approaches to optimization proposed [11][12][13][14][15], which concentrate around optimization of algorithmic component. However, this approaches have a common drawback: the process of optimization does not take into account exact properties of data being rendered.…”