On my current work in progress, I get this error message :"nct_getorientation: Error! Could not determine 3 different points!"
What is it ?
I also think that this is the reason why this model crashes Ayam on FreeBSD 5.3 (despite the fact that it works under win32 and Linux).
this error message is probably caused by a
degenerated NURBS curve. Ayam tries to calculate
the orientation of the curve in order to adapt
in e.g. an automatic trim curve generation but
fails to do so for your curve.
Do you care to share your model so that I can
have a look at it?
http://www.ayam3d.org/ Reconstruct the World!
Let's go share it !!
You'll find it there (http://frederic.praca.free.fr/3D/chevalAqsis.ay for the aqsis version and http://frederic.praca.free.fr/3D/cheval2.ay for the BMRT version).
Aha, thanks for that nice model. I do hope you
share the finished version with us!
The culprit lies in the "barre" object; it has
a much too big bevel radius, reduce this to
about 0.1. Other objects also have too big bevel
radii, e.g. the head of the horse. I had to
decrease it to "0.00625" to get cap surfaces!
Please be careful with caps and bevels, the
underlying algorithms are not bullet proof (yet),
and if strange surfaces result, Mesa may crash.
I could actually reproduce crashes with your
model, but they all lead to Mesa/SGI-GLU.
Don't worry, I'll share it !
The "barre" object (the stick). The bevel radius is equal to the radius of the ICurve because I wanted it to look like a kind of cigar. Reducing this to 0.1 makes it a beveled cylinder (which it is in fact). The other way is to do a concatenation of two quarter and a straight line and then to revolve it. But I found the strong beveled cylinder easier.
For the horse head, I don't really understand even if I've noted what you said. Such design was wanted because extusion was easier than any other design.
Did you notice the artifacts at both ends of what I called "bascule" ? Any idea about that ? I personnally think that this is a problem with my Concat object and the fillet option. What do you think about it ?
About the barre: You are right. But apart from
the bevel you requested caps from a curve which
is (due to the bevel radius) degenerated to one
single point in space (minus floating point
roundoff errors, which is why it seems to work
on some platforms)! So either switch off the caps
or decrease the bevel radius.
The horse head does not work right because of
the way the ICurve represents the starting point
of the internal NURBS curve. You can see the
error easily by converting the ICurve to a NCurve
"in place" and then zoom to the starting point.
The bevel intersects itself!
To get around this, just move the starting point
of your ICurve away from the corner to about the
middle of the lower straight part of the curve.
You will need to insert new points and delete
and move some other points to do this.
Check the bevel for self intersections!
Thanks, it seems to work correctly now, even if I still have some artifacts on the "bascule" objects.
The bascules concat curves with weights (the
half circles) with other curves, which does not
work well. I suggest to do the following:
Reset the weights of the half-circles (<W>), switch
their Knot-Type to "NURB", move the control points
so that the transitions to the long curves
are smooth and the "half-circle" is circular
again (requires some tweaking); finally, switch
ConcatNC Knot-Type to "Custom".
No more artifacts, thanks Randolf !
Log in to post a comment.