Particle-based Volume Rendering Research Articles

Extension of particle-based in-situ visualization for multipoint VR visualization

In the field of nuclear engineering, complex simulations on exascale supercomputers generate large-scale data. To facilitate efficient analysis of such simulation data, one needs to share them among scientists at remote locations. However, data I/O and data transfer for such large-scale data are quite costly. To resolve these issues, we developed a remote in-situ visualization system IS-PBVR based on the particle-based volume rendering (PBVR), which is suitable for parallel processing on modern supercomputers. In this study, we extend IS-PBVR for VR visualization on multiple client PCs, thus developing a multi-point remote VR visualization. We apply this technique to fluid simulations on GPU-based supercomputers and verify its utility by sharing in-situ VR visualization between multiple client PCs.

Interactive steering on in situ particle-based volume rendering framework

The development of supercomputers and multi-scale computational fluid dynamics (CFD) models based on adaptive mesh refinement (AMR) enabled fast, large-scale, and high fidelity CFD simulations. Interactive in situ steering is an effective tool for debugging, searching for optimal solutions, and analyzing inverse problems in such CFD simulations. We propose an interactive in situ steering framework for large-scale CFD simulations on GPU supercomputers. This framework employs in situ particle-based volume rendering (PBVR), in situ data sampling, and a file-based control that enables interactive and asynchronous communication of steering parameters, compressed visualization particle data, and sampled monitoring data between supercomputers and user PCs. The parallelized PBVR is processed on the host CPU to avoid interference with CFD simulations on the GPU. We apply the proposed framework to a real-time plume dispersion analysis code CityLBM, which computes the lattice Boltzmann method on the block AMR grid using GPU supercomputers. In the numerical experiment, we address an inverse problem to find a pollutant source from the observation data at monitoring points and demonstrate the effectiveness of the human-in-the-loop approach via the in situ steering framework.Graphical abstract

粒子ベースボリュームレンダリングによる大規模データの可視化技術

粒子ベースボリュームレンダリングによる大規模データの可視化技術

Time-Varying Volume Compression in Spatio-Temporal Domain

Data compression is always needed in large-scale time-varying volume visualization. In some recent application cases, the compression method is also required to provide a low-cost decompression process. In this paper, we propose a compression scheme for large-scale time-varying volume data using the spatio-temporal features. With this compression scheme, we are able to provide a proper compression ratio to satisfy many system environments (even a low-spec environment) by setting proper compression parameters. After the compression, we can also provide a low-cost and fast decompression process for the compressed data. Furthermore, we implement a specialized particle-based volume rendering (PBVR) [2] to achieve an accelerated rendering process for the decompressed data. As a result, we confirm the effectiveness of our compression scheme by applying it to the large-scale time-varying turbulent combustion data.

Remote Visualization System Based on Particle Based Volume Rendering

Remote Visualization System Based on Particle Based Volume Rendering

粒子ベースボリュームレンダリングを利用した遠隔可視化システムにおける粒子生成処理の高速化

複雑なシミュレーション結果を解析し,理解する上で,スーパーコンピュータ上の大規模データをボリュームレンダリングで可視化することが有効である.粒子ベースボリュームレンダリングを利用した遠隔可視化システムは,大規模なボリュームデータをサイズの小さい可視化用粒子データに変換することで従来手法のメモリ使用量,および,データ転送量の問題を解決し,対話的な遠隔可視化を可能とする.本研究ではスーパーコンピュータ上で動作するサーバ処理の負荷分散システムを構築し,京コンピュータ及びBX900を利用して約1000並列までのストロングスケーリングを達成し,一億要素のデータを数秒で処理した.

High-quality particle-based volume rendering for large-scale unstructured volume datasets

In this article, we propose a technique for improving the image quality of particle-based volume rendering (PBVR). A large-scale unstructured volume dataset often contains multiple sub-volumes, which cannot be ordered by visibility. PBVR can handle this type of volume dataset. Sampling misses often occur when the transfer function undergoes drastic changes, which can result in poor image quality. To reduce sampling misses caused by the high-frequency transfer function, we develop a new sampling technique called “layered sampling”. To confirm the effectiveness of our technique, we apply the proposed technique to a large-scale unstructured volume dataset subdivided into multiple sub-volumes.

Visualization of Large-scale CFD Simulation Results Using Distributed Particle-Based Volume Rendering

In this paper, we present a technique for visualizing a large-scale irregular volume dataset that is generated from an LES-based CFD simulation. Since our computational mesh was too large for a single computational node, it was divided into multiple regions; moreover, the resulting file was comprised of various irregular volume datasets. In order to cope with the multiple volume datasets, we extended our particle-based volume rendering (PBVR) technique so to fit the distributed computing environment. We applied our distributed PBVR technique to an LES-based CFD simulation, which explores dental fricative sound sources in order to confirm the effectiveness of the technique.

Improvement of particle-based volume rendering for visualizing irregular volume data sets

We present a technique for previewing large-scale irregular volume datasets using an improved particle-based volume rendering (PBVR) technique. Volume rendering of irregular grid volume data is considerably more complicated than that of regular grid data, since the sampling and compositing processes, which must be done in visibility order, are not straightforward. In our original PBVR, rendering involves generating and projecting sub-pixel-size, opaque, and emissive particles without visibility ordering. To make it easier to preview large-scale irregular volume datasets, we improve our original PBVR technique in two respects. The first is that we exploit its scalability to develop a cell-by-cell particle generation technique. The second is that we reduce the memory cost of the frame buffer using a pixel-by-pixel superimposing technique. To measure the effectiveness of our proposed method, we apply it to huge irregular volume datasets composed of 71 mega hexahedral cells or 1 giga tetrahedral cells.

Particle-based multiple irregular volume rendering on CUDA

In this paper, we describe an improved particle-based volume rendering (PBVR) technique for previewing a large irregular volume dataset using the CUDA architecture. This technique allows for opaque and emissive particles to render translucent volumes without visibility sorting. Our GPU acceleration of PBVR provides the multi-volume rendering feature while remaining compatible with both regular and irregular volumes. We also reduce the memory cost required for storing all sub-pixel values by proposing a pixel repetition technique for a large sub-pixel level. By adjusting the repetition level, we achieved a very smooth level of detail (LOD) control for trading quality for speed. Our work demonstrates a full-detail rendering rate from 5 to 10 fps for irregular volume data with mega-scale cell numbers on an NVIDIA GeForce 8800GTS.

“Effective Remote Visualization for Large-scale Volume Data by Particle-based Volume Rendering”

“Effective Remote Visualization for Large-scale Volume Data by Particle-based Volume Rendering”