Re: [ff3d-users] 'Segmentation fault' always?!

[Stephane Del Pino]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dear Arun,

I check your data. You did no mistake here.

Of course as I said previously this is due to the surface mesh generator that 
crashes. There are many reasons for that, but the most important, is that the 
marching tetrahedra algorithm may create very small triangles with bad aspect 
ratio. This leads to very unprecise normal evaluation. This generates a lot 
of problems that Cecile first tried to solve case by case. This was a very 
complex task and we finally came to the conclusion that if the algorithm was 
good on the paper, implementing it may not be possible because of numerical 
errors due to the use of computer algebra.

Using not to complex scene usually works because thanks to Cecile's good job, 
the most common cases have been isolated. Often, when it crashes, playing 
with the background mesh size changes the situation and may avoid not 
implemented cases.

But your particular scene is already too complex: since you are using 25 
cylinders, this requires the computation of the intersection of 75 primitives 
(3 primitives for each cylinder). This means that there are more chances to 
encounter a bad situation...

I tried your data and I have to admit that I see no solution there, and the 
new mesh generator that should hopefully fix that is far from being ready.

But you can still perform the computation with ff3d. For that you will have to 
forget about POV-Ray and built an unstructured mesh with an external tool. 
This implies some more work from your side but will work.

Having a look at
        tetgen  http://tetgen.berlios.de/
or      gmsh    http://www.geuz.org/gmsh/
may be a good idea.

Regards,
Stephane.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFBCB9ZbZyhKhZwdc0RAv5hAJ9S8HyLMySMKePc+ZElL4w+hCr+JgCeIq/9
dmXQ3OHq/UY47/DzeLCA11o=
=2/24
-----END PGP SIGNATURE-----