Visualization Software Packages Scott A Klasky PPPL September
Visualization Software Packages Scott A. Klasky PPPL September 25, 2000 Scott Klasky (PPPL)
Outline • • • VDE 2000, VDE 2001 What is visualization? Who cares? Criteria to judge a visualization package. Visualization Packages – Amira – AVS/Express – IDL – – – • • IRIS Explorer Open. DX PV 3 Sci. Run Vis. AD VTK Conclusions Demo Figure 10: Example of the Volume Visualization of two adjacent teeth; left: Vtk; right: Open. DX. The number of triangles in both surfaces are approximately equal. Figure 11: Examples of the Volume Visualization of a molar; left: VTK; right: Open. DX 9/9/2020 Scott Klasky (PPPL) 2
VDE 2000, VDE 2001 • VDE 2000: April 27 -28 2000 – – • Held at PPPL with 152 people in attendance http: //w 3. pppl. gov/vde 2000 Goal was to provide audience with knowledge of the popular Visualization Development Environments. The winners: Amira and Open. DX VDE 2001: April 26 – 27 2001 LBNL/NERSC Session 1 9 -2 Software vendor overview 2 -6 User talks 9 -11 11 -2 2 -4 4 -6 Session 1 9 -12 User talks 9 -10 1 -5 9/9/2020 9 -12 Poster Session Showdown Session 2 What's new/changes vendors University Vis. Tools Panel 1: Defining Scientific Visualization Benchmarks Panel 2: Extensions to Vis. packages for distributed memory environments Session 2 Hardware vendors Scott Klasky (PPPL) 3
What is visualization? • • Data Visualization is the process of representing data as a graphics display on a screen. Visualization gives you: – – – • Visualization tells the story of your data. – • Good stories capture people’s attention, and make them ask questions to gain new insights. Interaction with the data. – – • Insight into your data by allowing you to interact with your data. Enables you to analyze and obtain new information. Enables you to present your data with others. Visualization systems should allow us to interact with the data easily. Rotations/Translations/Zooms, Time animations, etc. must be FAST! Visualization Techniques – – – Isosurfaces Volume Rendering Slicing/Clipping Extraction (3 D->2 D->1 D) Stream Lines Figure 3. IRIS Explorer session using remote isosurface computation 9/9/2020 Scott Klasky (PPPL) 4
Who cares! • • • Back in the stone-age, researchers had X-terms which were incapable of scientific visualization. In the future (days to come) we will have PC’s with fast graphics cards! The world will seem like a different and better place – – • New CAD (Dell machines) has changed the way the engineers are working. 3 D vis. Packages will be available on your desktop! Your needs can be incorporated in what we do here for visualization! 9/9/2020 Scott Klasky (PPPL) 5
Criteria to judge a visualization package 1. Quality of the Output – Accuracy of the output • – 2. Isosurfaces, Streamlines Contour Plots, Volume Rendering Speed for rotations/translations/zooms. Ease of Use • • 4. 5. Aesthetic quality of the output Speed of Common Visualization Routines. • • – 3. Ex: There are many isosurface routines which miss triangles, Need statistical sample of users How many lines of code does it take to create everyday visualizations. Number of Visualization and numerical modules available. Cost 9/9/2020 Scott Klasky (PPPL) 6
AMIRA • • • PPPL is in the process of obtaining the license ($3000/node locked license) (Amira. Dev), $1 K for Amira 2. 1 (non extendible). At VDE 2000, they showed the most impress 3 d graphics. Data Flow environment, with less tools than DX, Express, but the tools it includes are more powerfull. – • • TCL command language for scripting or vis. Network editing. Extendible with the Amira. Dev edition (will be available in a another week). – • • 120 modules to visualize, analyze and process data. Extendible means that we can create new modules in the system. Will be primary used at PPPL for end visualizations (not as much for everyday visualization, unless it proves to be extremely powerful. Stereo Support 9/9/2020 Scott Klasky (PPPL) 7
AVS • • Not recommend by Muse/AVS if you have AVS/Express. Similar to IRIS Explorer and IBM DX. – – • • • All 3 products came on the market about the same time (AVS was the first). Does not have custom UI modules. More difficult to use c. f. Express. Does not support vector postscript output. Not supported at PPPL Stereo support Demo 9/9/2020 Scott Klasky (PPPL) 8
AVS/Express • • AVS/Express is a multi-platform, component-based software environment building applications with interactive visualization and graphics features We have 6 licenses (5 vis. editions), 1 developers edition). – – – • • • Solaris 2. 7 (for Express 5. x)…. Solaris 2. 6 (for express 4. 2), Dec Unix, Linux (wait for Express 5. 5 ->soon). 5 licenses for $5 K (Vis. Edition), $5 K for 1 developers license. Stereo graphics support Seems to be the most powerful visualization package here at PPPL. 2 x+ faster than DX for Xianzhu Tang’s datasets. Demo 9/9/2020 Scott Klasky (PPPL) 9
IDL • • • Great imaging software. Complex method of Visualization: Non-pipeline-dataflow program. Rendering Quality is not as good as other products. Very difficult to program for visualizaton: examples: VDE 2000 data, VDE 2000 data 2 Not considered a Vis. Product by Vis. Experts Demo 9/9/2020 Scott Klasky (PPPL) 10
IRIS Explorer • • • Best Quote from VDE 2000, “…explorer used to be everybody's favorite when DX was not free and SGI IRIS Explorer was, now we are in trouble…” Better than AVS, but still uses the same architecture. Memory is copied between modules. Has a collaborative interface. Stereo support Link to NAG math routines. 9/9/2020 Scott Klasky (PPPL) 11
Open. DX • • Similar to AVS Released to the open public May 18, 1999. – – • • Is well tested, widely used. SMP-thread based Support is continuing for this product from the mailing lists. Will be a PPPL supported Vis. Package. Very good data model Can run separate modules on separate computers. Does not support vector postscript output. DDX!!!! Demo 9/9/2020 Scott Klasky (PPPL) 12
PV 3 • • More of a university product by Bob Haimes (MIT). Built on top of PVM, and made to run on parallel computers. It co-processes the visualization of data generated on a parallel computer. Data generated from large scale scientific simulations can inflict huge burden on those tasked in assisting the scientist with understanding the results. – • PV 3 does not perform traditional scientific visualization on the data-set after it has been generated. – • Data from one simulation can occupy terabytes of data. Compress the data so it can be visualized. It deals with the volume of data concurrently with the solver. PV 3 works for parallel fluid simulations. – It contains all of the extraction tools used for these simulations including: • • • – • geometric cuts iso-surfaces Streamlines transient particle traces … These abstracts are stored on disk, and transferred to the vis. Machine. Not easy to set up! 9/9/2020 Scott Klasky (PPPL) 13
Sci. Run • Chris Johnson (Utah) is the creator of this code – – • Scientific Programming environment – – • • Is trying to create this is a commercial product. Is geared for medical researchers. Provides for the construction, debugging and steering of large scale computer simulations Extendible. Visual Programming Mainly used for “Steering”, where the simulation is built with tools in Scirun. – 9/9/2020 Runs on the Origin 2000, and other SGI’s. Scott Klasky (PPPL) 14
Vis. AD • Created by Bill Hibbard (Wisconsin). – – – • • • Mainly used for weather applications. Pure Java (Java 2) Uses Java 3 d and Java 2 d Fully collaborative. Fully platform independent (even runs on Macs…. Sort of->Os. X). Supports computational steering. Users must program in Java (or now Python) to get their vis. Not for serious 3 D visualization, but great for 2 d vis. , and some 3 d vis Has a nice Visualization Spread. Sheet program that lets you get nice visualizations without programming. • 9/9/2020 Scott Klasky (PPPL) 15
VTK (Kitware) • • Open. Source Used for 2 d Imaging, 3 d graphics and visualization. Distinguishing feature is the integration of functionality, support of supercomputing and activie community support. Designed as a toolkit (NOT A SYSTEM). – – – • • • Contains the most advanced visualization routines, although the other products are already incorporating these in their products. Contains streaming support, which allows VTK to handle large (Terabytes) data-sets. PPPL (with Stanley Dunn) worked on Parallel VTK with Kitware and Los Alamos. – – • • It is meant to be embedded into an application in combination with other tools. VTK’s core is in C++ Wrapped with TCL/TK, Java, Python. Results were poor, and we viewed this as difficult to use. To use, we must have the VTK manual by our side to know the interface of modules. Used mainly by vis. Experts. Demo 9/9/2020 Scott Klasky (PPPL) 16
Conclusions • PPPL supports the following vis. Packages. – – IDL: Bill Davis et. Al. AVS/Express • • • – – 9/9/2020 Easy to add modules Easy to program fast! AVS but users should work with Express, since we will not continue our license for this product, since Express can use AVS networks. Open. DX DDX: It’s Free, and it uses SMP->thread based, and it will eventually run on parallel clusters. VTK If there parallel implementation gets better (a complete rewrite of their system), and if they start to build a visual front -end, then we will look into this product again. Amira? ? ? If it can produce sexy pictures easily, then we should pay attention to this product…. Coming soon? ? ? If congress allows us (no money in M&S this year). Scott Klasky (PPPL) 17
- Slides: 17