andyrose@batcomputer.tn.cornell.edu (Andy Rose) (11/07/90)
This may give you an idea of what is up at Dept. Vis Cornell Theory Center. This was an application for credit for an independent study (undergrad type stuff). September 17, 1990 Methodology of Interactive Scientific Simulation II Visualization of Fluid Flow, Prof. Liu, Civil and Environmental Engineering Researchers in computationally intensive fields are often confronted with problems of data representation when producing results intended for analysis by humans. When vast amounts of numbers are computed in short order, more inventive ways of representing, displaying, and rendering data are needed. The users of the computing facilities at the Cornell National Supercomputing Facility have the capability of producing just such quantities of data, and the Department of Visualization is responsible for applying advanced techniques to meet this need. Because of recent advances in hardware it is now possible to simulate phenomenon while interacting with the computer, as opposed to specifying parameters before hand and running simulations unmonitered. Investigators are also provided tools to efficiently manipulate the display of data, providing control over color, point of view, lighting, and scale for creating displays with the most scientific merit. Because these applications require expensive equipment, little has been written about the merits of one visualization technique over another. I will work this semester with Prof. Liu, visualizing the results of inquiries into the nature of fluid flows. This type of data is common to many scientific investigations (time based lists of particle positions) and the work will have significant fall out for the Department of Visualization. Tools will be written to read arbitrary scientific data, and hopefully emerging standards for representing such data can be compared. Methods to display "fluid flow" data will be investigated, starting with displaying each point as a sphere, using existing software. Inquiries into the efficacy of other techniques such as isosurfaces, volume rendering, and particle advection will be made. A previous investigation into balancing visual complexity and interaction speed was completed last semester (Methodology of Interactive Scientific Simulation, March 8 1990, Prof. M. Salpeter) and results from that project will help guide this work. Information concerning acceptability of feedback speed for a researcher was gained, and this "reaction time" was found to be less important than previously thought. This could be because most researchers are unfamiliar with real time interaction, mostly gained from using quick computers (or video games). The main objectives of this study are: To find ways of representing general scientific data, especially in reference to existing standards such as the Hierarchical Data Format Vset from the National Center for Supercomputing Applications, U. Illinois, Champagne- Urbana. To develop tools for reading such data to conform to the specifications of the Advanced Visualization System from Stardent Computers. This software has been used successfully to visualize wide ranging types of data and indicates the direction that visualization is taking. AVS allows the researcher access to many tools which use standardized data formats and exist within in a user defined "network". The researcher is able to view his data in a number of ways in one session, rather than developing code for each visualization technique. To develop tools specifically for displaying "fluid flow" data and analyzing the value of these tools. Methods include plotting, rendering of spheres, isosurface display, and 3D vector display. To evolve a system in the Department of Visualization whereby a researcher can move from data generation to video tape output without the almost constant intervention and hand-holding now required. The main areas of investigation are in scheduling the resources of the video equipment and the graphic workstations, both of which are limited resources. I will be producing video animations for Liu and a paper examining these issues and others which may arise. -- Andrew Newkirk Rose '91 Department of Visualization CNSF/Theory Center 632 E & T Building, Hoy Road Ithaca, NY 14583 607 254 8686 andy@cornellf.tn.cornell.edu