In the attached I have a section of the most difficult part of one of my hulls to reproduce. I would like to fillet the edges of the indicates surfaces a 1". The two edges are 2" apart. At the aft end, as the sides flatten out, the radius needs to go to zero. The rolling ball method appears fit here. This works pretty well on the edge at the left (leaving a hole to be filled—yet why?).
However, the “ball” goes totally off the rails on the edge at the right. I am curious why that is the case. Decreasing the tolerance increases the chaos at the end. I would expect the ball to hit the end of the edge and stop, rather than curving around.
one of the edges still has a kink around 10 degree - the ball will not role to one point.
for the other edge:
a workflow how to get a precise pointy-fillet is described here: https://wiki.mcneel.com/rhino/tancylinders
in fast:
_filletSrf (Extend = No)
_split (Isocurve, Shrink = Yes) - optional step
_extendSrf
_setPt
But this does not do a nice job in your case.
I think building the extension with a _sweep2 should work.
the isocurve / CV s of the surface are quite dense…the wavy / bulky part might cause problems…
EDIT:
maybe something like this - at least for the edge that s nearly 0-degree at the end… Problem Fillet 3_tp.3dm (8.3 MB)