I can try to briefly answer some of these questions. I am sure others will respond with additional information. For your reference, I was involved with DX (and the former commercial group) from nearly the beginning having joined that group at IBM Research in early 1990. Among my interests were earth and space sciences, having done work mostly in upper atmospheric and space sciences at NASA/GSFC for the previous dozen years. This included work with a number of spacecraft-based data sets, including TOMS (Nimbus-7). You can see more about it from my web site, http://www.research.ibm.com/people/l/lloydt/. My primary focus these days is in applications of numerical weather prediction. I do some processing of GOES imagery within DX mostly for forecast verification. My other web site, http://www.research.ibm.com/weather, will have pointers to some of this work.
I think that the relevance of OpenDX to working with satellite data has been well-established. Admittedly, there are issues of scalability in dealing with large images. But that situation is likely worse with other packages. There is on-going work (e.g., by Greg) to address such problems on clusters of SMPs. 1. The traffic on the three e-mail groups for OpenDX seems to be pretty good. I don't change versions often or watch the bug fixes closely, so I can't comment on frequency. But it seems to be on-going at a reasonable rate. The mail groups seem to be effective in getting questions answered. Those of us that have used DX a lot try to help when we can. Although I admit to not always having time. But the best thing is that the number of people that have started to use DX and participate in the mail groups has grown significantly over the last year or so. So, I think for the help/community system works well. The software is being used in a number of "real-life" settings, at least seeing comments on the mail groups and knowing a little about DX users over the years. I'll let those users comment on their relevant work. Among other things, we are running a mesoscale weather model operationally nested to 1km resolution for the New York City area. This is tied to several interactive workstation applications (Win32, Linux and AIX) in our lab, and production visualization to populate a web site. I recently presented this work at a conference at ECMWF. I don't know if anyone from ESA was there last month. 2. Yes, that's true, compared to facilities in commercial packages like IDL or open source like gnuplot. The idea was to provide a basic facility to complement the more extensive visualization capabilities. After all to implement a more complete plotting package would be reinventing what already exists elsewhere and adding quite a bit of bulk to the system. Unless you intended the plotting to be interactive with custom direct manipulation within plotting windows, you could add something without doing any X/Motif. The easiest thing would be to write a module or macro that extracts relevant data and metadata and passes it to another package for better plotting. That package would do its own rendering and display/interaction. A next step would be to set up communication between that package and DX. Another approach would have a master program invoke both DX and another package from an integrated interface. All of these ideas have done by several groups over the years in either Unix and Windows environments. 3. Yes. Dave Thompson's company has a nice packaging for Win32. I have certainly gotten very good performance with 3d hardware on earlier releases, but admittedly with some fiddling with the configuration (e.g., NT). There has been extensive comments by several people on these e-mail groups about what to do, particularly by Suhaib. 4. This has been done, but I don't know its status. I'll let Greg or others to comment. Personally, I'm already moving more of my applications from Windows to Linux so this is of less interest. 5. Yes, that's right. I wrote one set eons ago. They aren't maintained primarily because it's not part of my current work. I try to answer questions about them. But you should be able to follow the approach to add or improve the capability. 6. That would be my understanding. 7. There have been some surveys and comparisons between OpenDX and VTK recently. Examples include a workshop on "modular visualization environments" sponsored by the Princeton Plasma Lab early last year. Others have been at tutorials, during the IEEE Visualization conferences. Briefly I think that OpenDX has advantages of a more comprehensive data model, visualization tools and interfaces (API, scripting and visual programming) and extensive utilization in earth/geographic applications. VTK benefits from being a newer (object-oriented) implementation since it is essentially an API. It's been open source for a lot longer than DX and thus, is popular in computer science groups during visualization. Another open source tool to look at is Vis-AD from U. Wisconsin. It doesn't have as much 3d capabilities as OpenDX or VTK. It is geared more toward analysis and processing, but has far better visualization tools than something like IDL. It is a essentially a high-level language written in Java (and leveraging the networking and other inherent facilities) with an elegant data model (less comprehensive in implementation than DX, but with a richer metadata implementation). It is being used by a number of research groups in atmospheric/earth sciences, which was a key goal in its design and implementation. I hope this helps. -------------------------- Lloyd A. Treinish Deep Computing Institute IBM Thomas J. Watson Research Center P. O. Box 218 Yorktown Heights, NY 10598 914-945-2770 (voice) 914-945-3434 (facsimile) [EMAIL PROTECTED] http://www.research.ibm.com/people/l/lloydt/ http://www.research.ibm.com/weather Sidney Cadot <[EMAIL PROTECTED]>@opendx.watson.ibm.com on 12/08/2001 05:28:35 AM Please respond to [email protected] Sent by: [EMAIL PROTECTED] To: [email protected] cc: [EMAIL PROTECTED] Subject: [opendx-general] assessing fitness of OpenDX for satellite data imaging Hi all, At the company I work for we are just starting a project to produce visualization software for the European Space Agency (ESA). This is going to be a sponsored open-source project so (naturally...) I'm looking at OpenDX to see if it addresses our needs. OpenDX seems to be a great tool but I've only been looking at it for a few days so I haven't delved in too deep yet. Our primary focus will be the processing of geographic/atmospheric data coming from the Envisat (to be launched sometime in 2002); it has three instruments to collect atmospheric data (Sciamachy, Gomos, Mipas). Furthermore, we will also be processing TOMS and GOME ozone data. Our goal is to have a multi-purpose, extensible tool for scientists in the ozone / atmospheric research area enabling them to analyze and compare the data coming from all those instruments. For our assesment the following questions are most important: (1) how active is the user/developer community, with regard to things like release frequency, bug report responses, and help with problems (I'll be a newbie for months to come, and it sure would be nice to know that there's a place to ask silly questions). Is OpenDX being used in any real-life commercial/scientific setting at the moment? (2) The 2D 'plot' facility seems rather limited in scope. I haven't been looking into the source code, but my guess is that we'd have to write a module providing somewhat fancier plots. Would this be difficult? I'm quite experienced at C-programming but I have no experience with X/Motif programming. (3) is it possible to get OpenDX running well on a Win32-based OS? Using Linux, opendx works like a charm for me, but my mileage on several win32 machines (win98, win2000, using exceed) seems to vary. For example, I can't get opendx to use hardware 3D rendering although I installed Exceed3D. (4) are any efforts under way to provide native win32 graphics support (...probably not, but just checking...) (5) there seem to be several add-on packages that provide mapping of geographical data onto a spherical earth; is any of those actively maintained? (6) Concerning licensing: my guess is that there cannot be a problem if we release our software as an "add-on" package to opendx (or even as a set of patches to the original, if the need would arise). Am I correct in assuming that? (7) How does OpenDX compare to other open source visualization packages? The only viable alternative I know of would be to use the Visual Toolkit (VTK), are there any others? Why should we use OpenDX (or why not?) Any info would be greatly appreciated! Regards, Sidney Cadot Science [&] Technology Delft, Netherlands
