160 likes | 304 Views
SV2. See me at the NPACI/Alliance booth for a live demo. See me at the CRPC booth for a live demo. Collaborative Scientific Data Visualization Framework. Ki, Klasky , Fox Syracuse University (NPAC) scott@npac.syr.edu. SV2 Features. Java2D (promise for fast 2D rendering/animations).
E N D
SV2 See me at the NPACI/Alliance booth for a live demo See me at the CRPC booth for a live demo Collaborative Scientific Data Visualization Framework Ki, Klasky, Fox Syracuse University (NPAC) scott@npac.syr.edu
SV2 Features • Java2D (promise for fast 2D rendering/animations). • Java3D (promise for fast 3D graphics on all platforms). • Voluminous data server (compression schemes to allow for fast data transfers). • Multiplexing (efficient collaboration data server along with TANGO)
OBJECTIVE • Our goal of this collaborative system is to use recent computing technology, Java, to build a multi-user collaborative scientific design and analysis environments which can run on all platforms. • The objective was to develop the scientific software environment where multiple users can create, share, manipulate, analyze, simulate, and visualize complex data sets over a heterogeneous network of PC’s, workstations and supercomputers.
Scivis->SV2 • Our SV2 research and design is based on the success of Scivis. Scivis is a collaborative scientific data visualization package written in Java. • Scivis allows users to visualize their data, which is piped in via sockets. • Although Scivis has been widely used in the “Binary Black Hole Grand Challenge”, it still had major limitations. • One of the most common complaints against Scivis was that users could only collaborate with other Scivis users. • SV2 is being designed such that users of Scivis3d (a Java3D version of Scivis), AVS, & VRML can collaborate together.
SV2 System Architecture Local Client Local Client Data Viewer 1 Data Viewer 1 Data Viewer 2 Data Viewer 2 Client Manager Client Manager Collaboration Tool Geometry Engine Filter Engine SV2 Server
Sample Screen Dump Surface & Contour Plots Raytracing Rotations of raytracing Isosurfaces X,y animations
USE JNI to incorporate quality visualization code. • The Stanford Volpack routines for raytracing. (written in C). • Isosurface rouintes with • We provide API’s too incorporate such routines into the SV2 server with minimal pain. • Eventually we will provide API’s to incorporate VTK filters too. (C++ routines) • Users can customize routines for their own use.
Future Work • Filter/Map creator, users can hook filters together, and create new API’s for those maps. Mysv3d(name,time,g3d,n) Myray(name,time,g3d,n) Downsize Downsize VTK:raytrace Smooth Isosurface Compression decimation Compression
SV2-Client Window • Data is sent from simulations, or files, to the SV2 server. • Data from the SV2 server is stored (in memory, and on disk). • Data file headers are sent to SV2-clients. • Clients can request to visualize the data. • To keep support for Scivis alive, we allow users to pipe data directly to SV2 clients.
SV2: Issues for rendering • For enhanced interactively, the image should be refined progressively as the data comes in from the remote server. • Image quality must be controlled as a function of the network's load and client's hardware setup. • In order to store and transmit large scale data sets,compression schemes have to be utilized. • In order to avoid full decompression, a sophisticated rendering method should carry out computations in compression domains at the client side. • For the above two issues, wavelet compression is an obvious choice in our system.
SV2: More issues • With upcoming network computers, the capabilities of a local client might be reduced significantly. • A data representation and rendering method is required, which will avoid the full expansion of the data in the clients memory. • Java-3D allows programmers to specify geometry using a binary geometry compression format. This compression format is used with APIs, and can be used both as a run-time in-memory format for describing geometry, as well as a storage and network format.
SV2: Visualization • Once Data-headers are on the client, users can select different methods to visualize the data. • For example, for a 3D data set, users can select either isosurface or raytracing. • Users can also select methods to filter this data, such as triangle decimation. • Clients request data from the server, the server performs the appropriate filter(s), and then sends back the geometry (or image, or actual x,(y,(z)) data) back to the client. • The client (Scivis3D,VRML(not yet implemented)) visualizes the data.