First off thanks for the reply.
Your correct that isosurface is not the tool I'm looking for, it is close,
but not it.
I currently do have that cut-plain idea. I have total of 4 windows with 3
being the 2D map of the three orthogonal views with user controls to move
them around independently.
So let me give a little background on the data and as to why I want
discrete colored cells in a 3D structure and you or someone else might have
a good idea as to how to do this better than I since I'm new to OpenDX.
Basically this data is coming from an intelligent agent model (think spiffy
game of life). Of primary importance is the over all configuration of
agents in 3 space. Secondly I want color coding based on the agent's last
selected rule. So red could mean the agent's choice was to move while blue
could mean it reproduced. There are of course more rules and colors I want
to display but it is worthless planing if I can't get OpenDX to display it
in that manor.
This is what I mean by isosurface being close but not quite right. It gives
me the overall shape of the agents in 3 space but missing colors.
This seems like it should be simple.
Any suggestions are welcomed.
Brendan
|---------+---------------------------------------->
| | Chris Pelkie |
| | <[EMAIL PROTECTED]> |
| | Sent by: |
| | [EMAIL PROTECTED]|
| | son.ibm.com |
| | |
| | |
| | 07/24/2003 07:53 AM |
| | Please respond to |
| | opendx-users |
|---------+---------------------------------------->
>-------------------------------------------------------------------------------------------------------------------|
|
|
| To: [email protected]
|
| cc:
|
| Subject: Re: [opendx-users] Isosurface with Colormap coloring?
|
>-------------------------------------------------------------------------------------------------------------------|
I would argue that Isosurface is an inappropriate technique to
visualize this data.
Since your cells are discretized to 2 values (0 ignored), how can you
have any meaningful Isosurface other than =7 or =3? And therefore, how
can a single surface adequately show the extent of a complete cell,
other than to show its boundary? In other words, Isosurface's triangles
inherently cut through your discrete cells, and could therefore cut at
any angle and anywhere along the length or breadth of the cell.
Therefore, none of these cuts fully represents the data within the
cell; all are equally insufficient representations. Therefore, the
technique 'lies' or misrepresents. And you don't have a CIA director to
fall back on.
So, just turn on or off the set of cells =7 (Include) to reveal those
=3 (or vice versa) and observe that 'sparse' matrix (ShowBoundary). It
will look like a Lego construction (sort of). I think that's the
closest you can come to illustrating how the iso-valued cells connect
spatially, which is essentially what Isosurface is doing.
Under continuum conditions, Isosurface of 3.8 or 5.25 would mean
something, and would be found by intersecting the triangles of the
Isosurface with the cell edges at newly created positions along the
cell edges, at the points where those 'iso' values are computed from
the cell's 'corner' data. Then, by definition, the entire Isosurface
has the same (iso) value (like 3.8) and colors the same. For a
variation, see one of the samples (I can't think of the name offhand,
but it has the Thundercloud data in it), that Maps the (continuous)
volumetric data onto an Isosurface. However, in your case, this will
not give the desired result (I'm pretty sure), since it will
undoubtedly create triangles (in the Isosurface) that have =7 at 1 or 2
vertices and =3 at the other(s), and you'll get a gruesome looking
color contour across the faces. (One way to minimize the ugliness is to
make a color map whose Saturation curve dips to white at 5, then set
Hue to blue=7 and red=3. Thus, you should get only shades of blue and
red fading to white.)
Something else to try is MaptoPlane which can deal with discretized
(connection-dependent) data/colors in the original sample volume (I did
it recently so I know this is true). You can slice this plane through
your volume any way you want and all slices will be representative,
since they'll all look like randomized checkerboards.
_______________________________
Chris Pelkie
Scientific Visualization Producer
622 Rhodes Hall, Cornell Theory Center
Ithaca, NY 14853
