Fillet e Shell problem (did you improve them in Rhino 8 wip?)

After watching a Plasticity video, I wanted to play some fillets and shells.
Rhino 7 works better than past versions, it manages to perform fillets (g1) that are not very large in diameter, even where the two cylinders intersect in tangency.

There is only one critical point (the one circled in red): it would be preferable to have a softer fillet at that point (that singularity is really not recommended).

I tried running a shell, emptying only the two integer cylindrical faces. At any tolerance the volume created is open and with various problems. These particular situations should be fixed (I don’t know how Rhino 8 Wip behaves).

An developed ad-hoc tool/plug in would be needed that only performs fillet/blend edges and that works interactively, with handles (as Plasticity does), it would be very convenient and immediate to see the preview of the blend. Rhino would need it. If it only focuses on solving small cases, with the current tools, we would get to version 20 to have robust and convincing fillets!
fillet e shell problem.3dm (2.1 MB)

That is the correct result. You are always going to get a pointy fillet in that situation. Although there are 2 possible pointy fillets in this case.
If you want to avoid the pointy fillet you need to make the concave fillets all the same size.
fillet e.3dm (433.0 KB)

Thanks Jim, that’s true.
Other than fillets, the shell doesn’t work. It should be improved in cases like this.

Generally, after doing complex fillets the shell almost never works as it should.

Shell won’t fail if the fillets are made accurately.
fillet e shell.3dm (886.2 KB)

The fillets are not accurate in your model. The pointy fillets in particular are from accurate.
Rhino has always had problems making accurate fillets and accurate intersection curves wherever surfaces become tangent on a model. If that defect in Rhino was fixed then shelling something like this should not be much of a problem.

2 Likes

you can shell with shrinkwrap in v8 and it’s a virtually never fail operation.

you can get a nurbs result as well.

Great tool. But the accuracy? Can it be equaled to that of a canonical shell?

Accuracy depends on poly count, more polys= more accuracy.

depending on what you are doing this may or may not be a huge issue.

Thanks! :wink: