today my students did an exercise about filleting.
the green surface was searched - it can easily build by _revolve.
but geometrically also a _filletSrf should find this surfaces.
but _filletSrf produces a big mess.
see screenshots and attached file
I would expect that the fillet-Algorithm does not produce such a mess - it should not be to difficult to find some conditions to stop the surface creation at a certain boundingbox, at a curvature, Knot-Densitiy, distance to surface … etc…
thanks for having a look at this, track it, solve it.
Hm… modern art just not your thing, I guess? Thanks, I’ll take a look. @Tom_P - it looks to me like the problem arises because that C-shaped thing is very close to but not quite a clean 180 degree revolve - so it gets messet up where the inputs come tangent to one another - but not quite.
If I replace that surface with a revolve, it all works but it does still get more complcated than is ideal at the tangent location
The bounding Box of the fillet should never be any larger than the bounding Box of the Center curve that was used to create the fillet inflated by the size of the fillet radius. So yes, checking that the fillet created by FilletSrf is valid is dead simple and fast if anyone cared to bother to do it.
The problem in this case is that the curvature of one of the surfaces used to create filletsrf is identical to the curvature of the fillet radius. Ideally yes, it should find the correct solution but that would involve a level of accuracy in terms of computing surface offsets and intersections that Rhino doesn’t have. But anyway it should not create a mess.
