Hi,

   within the next five minutes I sent this mail off I checked the code and
found
the message in triangle.c. My guess was as you now confirmed that this is due to

an error in the triangle texture interpolation. However I could not create
manually
a situation that created this error. I have run an example with a triangle, that
worked
fine, then I made two vertices the same, still every thing okay ( no error
message ) but
of course no object was drawn, not even a line and finally I made all points the
same
:again no complaints. So I don't know what the problem actually is. But it seems
it
is save to ignore for the moment.

About my second question I found out that I suffered a tremendous performance
hit
when I use +MV3.0 switch ! Megapov is just 5-10 times slower ( rendering ). Does
any
one know why ? BTW povray 3.1 does not get slower at all - so it would seem.
( Not that is really important but puzzling if you are trying to use older style
code )
Thanks,
   Lothar


Nicolas Calimet wrote:

>         Hi Lothar  ;o)
>
> > all determinants too small
>
>         As you mention later in your message, this is a problem related
> to mesh2 and triangles that carry a texture at each vertex. This question
> has been asked by Warp in this group [message of 2001-06-28] and I posted
> an answer - a bit late actually.
>         The code was designed for interpolating 3 colors between the corners
> of a triangle, as written by A. Philippsen (you might know him: author of
> the DINO software). Later, Nathan Kopp improved the code to get full texture
> interpolation instead of colors only. But the code is rather not optimized,
> and some times you have a small problems that give this message. This most
> probably happen with "degenerate" triangles (part of them include triangles
> that have 2 or 3 identical vertices). IMHO this message can thus be simply
> ignored, it won't affect rendering.
>
> > In addition to that it seems very slow. I don't know whether this is a
> > related error or whether mesh2 can get slow.
>
>         On my pov4grasp page I did extensive benchmarks using mesh2 and
> interpolated-color-textures in mesh2 objects and did not notice particular
> difference in speed (as compared to original mesh object + Chris Colefax's
> macros, or my custom mesh object). I'm talking about rendering speed of
> course, not parsing. See http://pov4grasp.free.fr/features/grasp_surface.php3
>
>         For more discussions, don't hesitate to contact me by normal mail
> (my adress at work is prefered).
>
> *** Nicolas Calimet
> *** http://pov4grasp.free.fr

--
Dr. Lothar Esser
NIH / NCI
Tel. 301-435-6163
email les### [at] helixnihgov

Post a reply to this message