Efficient Data Visualization

Efficient Data Visualization is a research and development project for large-scale data, mainly in the field of computational science, to develop intelligent visualization techniques that do not miss important changes in the data. In this research, we are trying to solve various problems related to numerical data, which are becoming larger and more complex as supercomputers become more powerful and numerical simulation techniques become more accurate. The following is an overview of some of the research we have conducted to date.

Smart in-situ visualization for large-scale numerical simulations

In-situ processing has widely been recognized as an effective approach for the visualization and analysis of large-scale simulation outputs from modern HPC systems. However, traditional batch- based in-situ visualization can produce large amounts of rendering results for post-hoc visual analysis, which can make it difficult to gain rapid insight into the simulation results during post-hoc visual analysis. As an alternative to accelerate the process of obtaining scientific knowledge, we have worked on a smart visualization approach, focusing on extracting a set of images that may facilitate the rapid understanding of the underlying simulated phenomena. In this work, we present a method for automatically adjusting the camera focus point and zoom level during in-situ visualization. We integrated the proposed approach with the existing in-situ smooth camera path estimation method for evaluation purposes and used two CFD simulation codes and two HPC systems (x86 Server and Arm-based Fugaku supercomputer) for the evaluations. The preliminary results seem promising, and we are planning further improvements by working with domain expert collaborators.

Related papers

Taisei Matsushima, Ken Iwata, Naohisa Sakamoto, Jorji Nonaka, Chongke Bi, Information Entropy-based Camera Focus Point and Zoom Level Adjustment for Smart In-Situ Visualization, International Conference on High Performance Computing in Asia-Pacific Region (HPC Asia), pp.164-173, 2024
Ken Iwata, Naohisa Sakamoto, Jorji Nonaka, Chongke Bi, Information Entropy-based Camera Path Estimation for In-Situ Visualization, arXiv:2301.11591, 11 pages, 2023

Multi-kernel denosing autoencoder for particle-based rendering

Volume rendering has been an important part of scientific data visualization in simulations of medical anatomy and analysis of internal physical quantities. Reducing the computational cost and time consumption of rendering complex volumes using this rendering method is one of key challenges for many data visualization researchers. Particle-based rendering (PBR) method has been proposed as one of the volume rendering methods. PBR is an efficient method, which enables to render semi-transparency for both of geometric and volumetric objects without visibility ordering. In this method, the objects can be rendered repeatedly based on stochastic process, and then the final image can be obtained by ensemble averaging the rendered images. The higher the number of repetitions, the higher the render quality. However, this can be a time-consuming process and a major cause of performance bottleneck, especially in the case of large-scale and complex dataset. To overcome these problems, we propose Multi-kernel denoising autoencoder (MkDAE) to denoise objects rendered with low repetitions to obtain high quality images as an alternative to rendering with high repetitions, reducing the time consumption.

Related papers

Masahiro Morii, Amirul Haziq bin Azman, Naohisa Sakamoto, Takuma Kawamura, Multi-kernel Denoising Autoencoder for Particle-based Rendering, The 8th IIEEJ International Conference on Image Electronics and Visual Computing (IEVC 2024), 2024.3 (accepted)

Edge enhancement for particle-based rendering with ambient occlusion effects

With the increasing performance of the computing environment and the advancement of computer science, numerical simulation data are becoming larger and more complex. When visualizing these data, the internal structure is hidden by external objects. Semi-transparent visualization allows us to see the hidden structure, but it is difficult to perceive the shape and three-dimensional structure of the data. In this study, we apply ambient occlusion and edge enhancement to particle-based rendering. Particle-based rendering is a semi-transparent visualization technique that does not require depth sorting. Ambient occlusion intensifies shadows such as wrinkles and hollows, while edge enhancement makes the shape of semi-transparent objects more perceptible by increasing the opacity of the object’s silhouette. By adding these two effects to semi-transparent visualization, the three-dimensional structure of translucent objects can be easily perceived. In this study, we confirm that the proposed method can efficiently visualize complex structure by visualizing magnetohydrodynamic simulation data.

Related papers

Yasuyuki Fujita, Naohisa Sakamoto, Ambient Occulusion for Semi-transparent Streamlines with Stochastic Rendering Technique, Proceedings of the 47th visualization symposium (The visualization Society of Japan), 6 pages, 2019 (in Japanese)
Yasuyuki Fujita, Naohisa Sakamoto, Koji Koyamada, Ambient Occulusion for Semi-transparent Streamlines with Stochastic Rendering Technique, The 15th Asia Symposium on Visualization (ASV15), Abstract files (ASV-0205), 2019
Daimon Aoi, Kyoko Hasegawa, Liang Li, Yuichi Sakano, Naohisa Sakamoto, Satoshi Tanaka, Edge highlighting of laser-scanned point clouds improves the accuracy of perceived depth in transparent multi-view 3D visualizations, International Journal of Modeling, Simulation, and Scientific Computing, 2450021 (19 pages), 2024

Parallel particle-based volume rendering without visibility ordering

Large-scale simulations have widely been conducted on modern High-Performance Computing (HPC) systems in a variety of scientific and engineering fields, and scientific visualization has been a popular approach for analyzing and extracting meaningful information from the simulation results. In this work, we focused on Particle Based Volume Rendering (PBVR) method because of its proven effectiveness for handling non-trivial unstructured volume data, which is still commonly used on numerical simulations in the engineering fields. PBVR possesses a visibility sorting-free characteristics thanks to its use of small and opaque particles as the rendering primitives. However, there is a memory cost and image quality trade-off because of the necessity of storing the entire sets of generated particle data before starting the rendering process. In this paper, we present a memory cost efficient parallel PBVR approach for enabling high-quality and high-resolution PBVR of large-scale numerical simulation results. For this purpose, we focused on the image data gathering and processing instead of traditional particle data gathering and processing by using the sort-last parallel image composition approach. We evaluated its effectiveness on the K computer by using the Binary-Swap-based 234Compositor library, and verified its potential for reducing the memory cost while generating high-quality and high-resolution image data.

Related papers

Naohisa Sakamoto, Hiroshi Kuwano, Takuma Kawamura, Koji Koyamada, Kazunori Nozaki, Visualization of Large-scale CFD Simulation Results Using Distributed Particle-Based Volume Rendering, International Journal of Emerging Multidisciplinary Fluid Sciences, Vol.2, No.2, pp.73-86, 2010
Yoshiaki Yamaoka, Kengo Hayashi, Naohisa Sakamoto, Jorji Nonaka, A Memory Efficient Image Composition-based Parallel Particle Based Volume Rendering, Journal of Advanced Simulation in Science and Engineering (JASSE), Vol.6, No.1, pp.1-10, 2019

Order-independent semi-transparent rendering

We propose a technique for semi-transparent rendering, which can integrally handle irregular volumes and polygons without visibility sorting. A projection technique for rendering of the irregular volumes requires a large memory space to calculate the visibility sorting at each viewing point. To solve the problem, we regard the brightness equation as the expected value of the luminosity of a sampling point along a viewing ray, and we propose a sorting-free approach that simply controls the fragment rendering by using the evaluated opacity value to calculate a rendered image. In our experiments, we applied our technique to several numerical simulation results and confirmed its effectiveness by rendering the volume with its semi-transparent boundary polygons and demonstrating the application of our technique to a high-resolution distributed display system.

Related papers

N. Sakamoto, J. Nonaka, K. Koyamada, S. Tanaka, Particle-based Volume Rendering, Proceedings of Asia-Pacific Symposium on Visualization (APVIS 2007), pp.129-132, 2007
Naohisa Sakamoto, Takuma Kawamura, Koji Koyamada, Improvement of particle-based volume rendering for visualizing irregular volume data sets, Computers & Graphics, Vol.34, No.1, pp.34-42, 2010
Sakamoto.N, Koyamada.K, Stochastic Approach for Integrated Rendering of Volumes and Semi-transparent Surfaces, Proceedings of the 2012 SC Companion: High Performance Computing, Networking Storage and Analysis (Workshop on Ultrascale Visualization), pp.176-185, 2012
Koji Koyamada and Naohisa Sakamoto, Particle-Based Fused Rendering, High Performance Parallel Computing, Chapter 5, IntechOpen, pp. 55-69, 2018