So it seems to me that the only trick in that case is inverting to column major, which is accomplished as I described. You may change the delta vector to accurately reflect your actual deltas. In fact, you can induce skew with deltas such as:
delta 0 2.3 1.2
delta 0 1 0
delta 10.0 0 0
This simultaneously converts to X-fastest, but causes the X-axis to rise in Y by 2.3 per position as it moves right by 1.2 per position. Z meanwhile varies by 10.0 per position.
Nevertheless, the positions when laid out, are still connected by 'cubes' (in DX-speak), in this case 'improper parallelograms'.
All that said, if your deltas in X vary, say 1.2, 2.5,1.0,2.3....
then you need to look at "product arrays" section of the Native DX File Format. You write out the list (array) of all X deltas as one 'term' of the product array, then those of Y and Z, then in effect "multiply" by combining all 3 terms in a product array. 'Semi-regular' positions, still connectable by regular gridconnections. If any term is fully regular, you can use the shorthand origin, delta for it's term declaration.
If all else fails, create a net that does nothing but use General Format to Import the positions (reordering them as needed), then Export them to .dx format. Then do the 'data'. Then you'll have two .dx files that can be combined by a 3rd .dx file that contains only header info and refers to the contents of the other two files, by component names within the files. You can save the two 'converter' nets and run them in script mode to automate this whole process if you need to do it often. Your batch script can easily create the 3rd file on the fly once you have the template figured out.
On Tuesday, Jun 24, 2003, at 09:59 America/New_York, Maxim Loginov wrote:
Hi Chris_______________________________
there is little misunderstanding... from very begining:
I have plot3d grid file, stepsize of the the grid varies along every direcrion
(irregular grid in OpenDX notion, but not scattered points, just every
gridcell is improper parallelogram), gridpoints in it saved like (Fortran):
write (nfile) (((x(i,j,k),i=1,npx),j=1,1),k=1,npz),
+ (((y(i,j,k),i=1,npx),j=1,1),k=1,npz),
+ (((z(i,j,k),i=1,npx),j=1,1),k=1,npz)
so that first index varies fast. now I want to read this grid into OpenDX
using Native File Format. I have to specify order of indexes for proper
reading this array into OpenDX with Native File Format. but how? in General
Array Importer I can, but it is impossible there to read grid from one file
and data from another (as it done in plot3d format).
actually I do not understand how to use your advice: it is for regular grid,
where stepsize is constant along the direction.
best
Maxim
http://opendx.npaci.edu/docs/html/pages/usrgu025.htm#HDRARRAYS
The 'trick' (and it is somewhat obscure) is revealed in Figure 11 on
this URL.
Rearrange the delta clauses.
If Z varies fastest, you might write:
delta 1 0 0 # for X
delta 0 1 0 # for Y
delta 0 0 1 # for Z
But if X varies fastest, write:
delta 0 0 1 # for X
delta 0 1 0 # for Y
delta 1 0 0 # for Z
On Tuesday, Jun 24, 2003, at 04:45 America/New_York, Maxim Loginov
wrote:
Hi all
from OpenDX User Guide (appendix "B.2 Data Explorer Native Files",
paragraph
"Syntax of the Native File Format", subparagraph "Data section"):
The Data Explorer file format is flexible enough to describe many
existing
data formats without having to reformat the data. It allows you to
specify
byte order, which index varies fastest, whether the data type is
floating
point or byte, and whether the file format is binary or ASCII.
but I have not found how to specify _which_index_varies_fastest_. is it
possible at all in Native File Format? or I have to use General Array
Importer? last is not convinient, because my grid and field data in
different
files (actually I visualize Plot3D data and do not want to use
converter)
best regards
Maxim
--
Loginov Maxim
Institute of Fluid Mechanics
Department of Mechanical Engineering
Dresden University of Technology
01062 Dresden, Germany
ph.: +49(0)351 463 34910
fax: +49(0)351 463 35246
email: [EMAIL PROTECTED]
_______________________________
Chris Pelkie
Scientific Visualization Producer
622 Rhodes Hall, Cornell Theory Center
Ithaca, NY 14853
Chris Pelkie
Scientific Visualization Producer
622 Rhodes Hall, Cornell Theory Center
Ithaca, NY 14853
