D205 非定常大規模データのインタラクティブ可視化システムの開発

Translated title of the contribution: Development of Interactive Visualization System for Unsteady Large-Scale Data Set

小野 謙二, 玉木 剛, 吉川 広幸, 片野 康生, Kwan-Liu Ma

Research output: Contribution to journalArticle

Abstract

A parallel volume rendering system that incorporated a hardware compositor was developed to visualize unsteady large-scale data set in real-time. This rendering system is constructed by PC-cluster on a master-slave basis, and utilizes hardware functions of GPU on a modern graphics card. At first, a master node divides a data region into several sub-domains, and then each slave node renders the images for each domain. Generated images on each Slave-PC are gathered and overlapped each other according to a BSP tree, which is determined by the relation between an eye point and the position of each domain. This superimpose process is carried out on a composite hardware to minimize the transmission and synthesis time of the images. It was found that the present visualization system achieved enough interactivity with 10-15 fps for time varying data of 25.6GB.
Original languageJapanese
Pages (from-to)311-314
Number of pages4
Journal可視化情報学会誌 = Journal of the Visualization Society of Japan
Volume24
Issue number1
DOIs
Publication statusPublished - 2004

Fingerprint

Visualization
Hardware
Volume rendering
Composite materials
Graphics processing unit

Cite this

D205 非定常大規模データのインタラクティブ可視化システムの開発. / 小野謙二; 玉木剛; 吉川広幸; 片野康生; Ma, Kwan-Liu.

In: 可視化情報学会誌 = Journal of the Visualization Society of Japan, Vol. 24, No. 1, 2004, p. 311-314.

Research output: Contribution to journalArticle

@article{47700aba827d41239bf67e75b2c6c598,
title = "D205 非定常大規模データのインタラクティブ可視化システムの開発",
abstract = "A parallel volume rendering system that incorporated a hardware compositor was developed to visualize unsteady large-scale data set in real-time. This rendering system is constructed by PC-cluster on a master-slave basis, and utilizes hardware functions of GPU on a modern graphics card. At first, a master node divides a data region into several sub-domains, and then each slave node renders the images for each domain. Generated images on each Slave-PC are gathered and overlapped each other according to a BSP tree, which is determined by the relation between an eye point and the position of each domain. This superimpose process is carried out on a composite hardware to minimize the transmission and synthesis time of the images. It was found that the present visualization system achieved enough interactivity with 10-15 fps for time varying data of 25.6GB.",
author = "謙二 小野 and 剛 玉木 and 広幸 吉川 and 康生 片野 and Kwan-Liu Ma",
year = "2004",
doi = "10.3154/jvs.24.Supplement1_311",
language = "Japanese",
volume = "24",
pages = "311--314",
journal = "可視化情報学会誌 = Journal of the Visualization Society of Japan",
issn = "0916-4731",
publisher = "The Visualization Society of Japan",
number = "1",

}

TY - JOUR

T1 - D205 非定常大規模データのインタラクティブ可視化システムの開発

AU - 小野, 謙二

AU - 玉木, 剛

AU - 吉川, 広幸

AU - 片野, 康生

AU - Ma, Kwan-Liu

PY - 2004

Y1 - 2004

N2 - A parallel volume rendering system that incorporated a hardware compositor was developed to visualize unsteady large-scale data set in real-time. This rendering system is constructed by PC-cluster on a master-slave basis, and utilizes hardware functions of GPU on a modern graphics card. At first, a master node divides a data region into several sub-domains, and then each slave node renders the images for each domain. Generated images on each Slave-PC are gathered and overlapped each other according to a BSP tree, which is determined by the relation between an eye point and the position of each domain. This superimpose process is carried out on a composite hardware to minimize the transmission and synthesis time of the images. It was found that the present visualization system achieved enough interactivity with 10-15 fps for time varying data of 25.6GB.

AB - A parallel volume rendering system that incorporated a hardware compositor was developed to visualize unsteady large-scale data set in real-time. This rendering system is constructed by PC-cluster on a master-slave basis, and utilizes hardware functions of GPU on a modern graphics card. At first, a master node divides a data region into several sub-domains, and then each slave node renders the images for each domain. Generated images on each Slave-PC are gathered and overlapped each other according to a BSP tree, which is determined by the relation between an eye point and the position of each domain. This superimpose process is carried out on a composite hardware to minimize the transmission and synthesis time of the images. It was found that the present visualization system achieved enough interactivity with 10-15 fps for time varying data of 25.6GB.

U2 - 10.3154/jvs.24.Supplement1_311

DO - 10.3154/jvs.24.Supplement1_311

M3 - 記事

VL - 24

SP - 311

EP - 314

JO - 可視化情報学会誌 = Journal of the Visualization Society of Japan

JF - 可視化情報学会誌 = Journal of the Visualization Society of Japan

SN - 0916-4731

IS - 1

ER -