Personally, I tend to prefer lots of small modular files. However, these need not be standalone files if they share the same grid and/or positions: those can be 1 or 2 additional files.
So you can easily create a field_01.dx file that references a field_positions.dx and a field_connections.dx (common to all time steps) and that field_01.dx file can directly contain as many "data'-like components as you need, attached to those pos and conns. Then you end up with lots of individual fields which can be easily assembled by one series.dx file, which consists of nothing more than references to field_01.dx, field_02.dx and so on. This also permits making quick test series that contain only the first 2-3 files or every 10th file or whatever. In extremely large time series, this is very convenient!
Not sure what other problems you are having: do you have 3D positions that appear to lie on the sphere? If not, use your lat-lon values and deform them from rectangular to spherical polar: the 2D mesh will be transformed along with the positions. If you have many vectors along +-90 lat, you'll get a mess at the poles. You might need to average their values, then use invalid positions to eliminate all but 1 for clarity in the vector glyph plot you propose.
On Aug 25, 2005, at 9:52 AM, Hans Georg Krauthaeuser wrote:
Hey,<x-tad-bigger>_______________________________
After about 12 years of abstinence, I restarted to use opendx (I think
it was DataExplorer when I used it first). So my skills need a
refreshment...
This is my problem:
I have 3D vector data (current density) located at some (26) points on
the surface of a sphere. Longitude and latitude spacing are constant.
Poles are included. The current data depends on two other parameters
(frequency (type float): up to 2000 values, t (type int): 60 values).
What I wanted to get is:
- a 3D view of the sphere with the current vectors
- a 3D view of the magnitude of the current with smooth interpolation
- 2D projection of both
- possibility to chance the two parameters and redraw the images
I think, I can do the plots when I have loaded the data. (Nevertheless,
pointers to examples or hints are very much appreciated).
My question is:
What is the best way to organize the data?
A lot of small dx-files? One monster file?
How do I make a series with more than one parameter? With only the
frequency I would write:
object 1 class array type float rank 1 shape 3 items 26 data follows
...
object 2 class array type float rank 1 shape 3 items 26 data follows
...
object 1603 class field
component "positions" value 1
component "data" value 2
...
object "series" class series
member 0 value 1603 position 1e+007
...
Regards
Hans Georg Krauthaeuser
--
Otto-von-Guericke-Universitaet Magdeburg
IGET | fon: +49 391 67 12195
Postfach 4120 | fax: +49 391 67 11236
39016 Magdeburg | email: [EMAIL PROTECTED]
Germany | www: www.uni-magdeburg.de/krauthae
Chris Pelkie
Scientific Visualization Producer
622 Rhodes Hall, Cornell Theory Center
Ithaca, NY 14853 (607) 254-8794</x-tad-bigger>
