Hi,
I have a CT-scan of a rock core. I have been having a ball using
OpenDX to work
with this data and have been able to do just about everything I could possibly
want. However, I want to render the volume so that it looks like a solid piece
of rock. OpenDX's volume rendering module uses a transmitted light model, my
rock core looks ghostly. How can I get it to look "real"?
I don't think upping the opacity multiplier will do it. I guess MAP is the way
to go but how do I create a cylinder with closed ends to map to? I have tried
using Construct and compute to create three grids - a cylinder and two circles
that I collect and export - it all looks like a closed cylinder - but I am
having trouble getting map to work with it. Any suggestions would be more than
welcome.
If you've got that far, you're almost there. Don't Collect before
Map. Collect after Mapping the 3 pieces. Again (cause I know I said
this to someone in the last week): Collect makes a generic "group".
Generic groups can hold any pile o' junk, but many modules do not
carry around all the necessary overhead to sort through the pile.
Thus, there are more structured group types (multigrid and series)
that force members to be similar in some important ways; some modules
can then happily iterate over all members knowing in advance what
they are going to have to do to each. But in your case, don't even
bother: just Map onto the various parts, then Collect (->group of
weird stuff) and show the result (won't have nice seams, but hey,
it's free!).
Probably simpler than anything else I can think of... though I
considered suggesting you create (Compute) another data value that
describes an "isovalue" corresponding to your "closed cylinder" (you
do the math: think superquadric: after all, a cylinder is just a
sphere with corners), then after Isosurface yields the desired shell,
you Mark the "real" data and colormap or whatever.
This is more parsimonious than another suggestion that you cull out
the invalid rectangle filling values to be left with only the outer
rock core, then ShowBoundary as David suggested. The problem with
this approach is that culling (use Include) makes your regular volume
irregular, so it bloats up in memory big time.
The Isosurface value algorithm could be adjusted parametrically
(think radius control) to give different cylindrical "slices". Or
with more effort, you could map onto a dodecahedron or something.
have fun.
Chris Pelkie
Vice President/Scientific Visualization Producer
Conceptual Reality Presentations, Inc.
30 West Meadow Drive
Ithaca, NY 14850
[EMAIL PROTECTED]
