Still helping Fluent get their converter to OpenDX worked out and
encountered a strange phenomenon. Any insightful comments welcome:

They were supplying me with test sets made up of tetrahedra, that I was
visualizing in DX. I used "front colors" and "back colors" and normal
glyphs to check for the correctness of tet winding.

In a test case, I had 1 cube cut into 6 tets: 4 were wound correctly (front
colors were out and normals started on face surfaces and projected out) and
2 incorrectly (back colors showed and normals went through the interior of
the tet volume). Trying to simplify things for debugging, I modified the
.dx file to figure out which 2 were the bad boys so I could study their
connection list order.

Lo and behold, by eliminating (commenting out) all the "good" tets, the
previously "bad" tets became good (some sort of epiphany, I guess, or maybe
a relativistic moral upbringing).

Turns out that DX decides which way your tets are wound (ouch!) depends on
which one it sees first in the input list. That is, if I now add back in
one of the "good" tets to the list but position it AFTER one of the "bad"
tets, the "bad" one is good and the "good" one is bad.

Aiie!! The documentation clearly implies that one of the (2) correct ways
to wind a tet is to traverse the first face COUNTERclockwise, then go BACK
to pick up the 4th vertex. My description of a "good" tet follows this
convention, and in my worldview a "bad" tet traverses the first face
CLOCKwise, then goes BACK to 4.

But now my faith is shaken. Any apologists out there to counter my apostasy?

Chris Pelkie
Vice President/Scientific Visualization Producer
Conceptual Reality Presentations, Inc.
30 West Meadow Drive
Ithaca, NY 14850
[EMAIL PROTECTED]

Reply via email to