Another beginners question about _filletedge

Hello There. I am Michael from Switzerland, have some Rhino experience and this ist my first post.

I am trying to learn more about the challenges _filletedge seems to have. I came across this very “easy” task to fillet the red box, so it turns out like the closed solid in the middle. (The solid on the right side was made by me but without filleting). but it always turns out very bad. i reckon it has something to do with the factof the filleting radius 10mm being exactly half of the boxes width 20mm.
i searched several hours for an answer. is there a tutorial at mcneel’s adressing this sort of problems? i myself was not able to find anything like it. i’ll be happy to read me trough it first.

of course any other help would be most helfpul. thanks a lot.


fillet_basics_02_2020_v2.3dm (424.2 KB)

The problem is exactly that.
Rhino follows “rolls the ball” on the surfaces being filleted, so the fillet radius must stay smaller than the surface width or the entire original surface is consumed, and that is not allowed.

This Wiki page is getting old but the concepts are correct.:

For that shape, Fillet is the wrong tool.
Sweep an arc down the surface edges to make the semicircular surface.

The “mess” you see is how far the Fillet command made it before it ran off the surfaces and stopped with what it had calculated up to that point.

thanks john
so i red all these information about filleting. in this cas i suppose this can be applied:
“The radiuses used can not be so large that they overlap each other and completely consume any surface they are following.” (found in mcneel wiki)
do you agree?

Hi Michael - FilletEdge does not know what to do if the fillet needs to be attached to a face that is not one for which an edge was selected.


thanks pascal. i am not sure if i understand what you mean. filletedge fails if it comes across a face, whig is not in contact with the selected edge?

Hi Michael - I wouls start with this:

Then Trim, or Boolean operations to get here:

Then FilletEdge or, more glass-like, maybe, BlendEdge, to finish:




This is my version of the bottle design.
bottle.3dm (303.9 KB)

The trick to getting the 10mm fillets to line up as shown in the drawing. is to make the conic blue curve using the green construction lines. The arching fillet is made with filletsrf.

If you want to play with the shape of the arching fillet you could make the revolve surface from the blue curve with history and the arching fillet with history. Then by editing the weight of the center cv or by editing the degree of the blue curve you might be able to change the shape of the arch to better match the shape shown in the drawing.
Just make sure the end tangent direction of the blue curve remains unchanged

1 Like

I apologize for my late answer. Thanks a lot for your explenations and the file. i do understand and was able to reconstruct pascals bottle.

i failed tought to make jims one. how do you manage to get the purple curve exactly trough the edge of the fillet? (which i made in my model with a sweep1) i think the problem lies somwhere in the geometry bevore revolving (?)


Grappa compilation.3dm (1.8 MB)

Dear @Michael_CH
happy to see you here in the forum.
we will discuss this exercise on our Thursday’s lecture.
…so please be patient…
I will post the final solution the days after - similar to jim 's approach.
see you the day after tomorrow
kind regards


hi tom. thanks. i am looking forward to that. that bottle has something of a gordian knot. I wonder what Alexander the great would have done…

See you



1 Like

this is the original bottle that inspired this exercise…

grappa2023_03_3versions.3dm (664.4 KB)

the file above shows 2 solutions from yesterdays lecture.

the really sophisticated is the second bottle - that uses part of a sphere - so the bow like / arching fillet is actually a revolve.

thanks at @pascal @jim @John_Brock

kind regards - tom

1 Like

I assume by “purple curve” you mean the blue curve. Blue is the color Rhino says it is (your monitor may vary).

I explained in my previous post that the blue curve is a conic curve (part of conic section of a parabola).
The green curves are the construction lines needed to create the blue curve.

The red curveA is poorly defined in the 2D plan since it is a fillet connecting two curves that are completely undefined.
But anyway, if we assume the red fillet is in the correct place then green curveB is made tangent to the red curveA and the vertical green curveC is made tangent to the vertical fillet. To get it tangent to the vertical fillet I extracted an isocurve from one of the vertical 10mm fillet surfaces and rotated that 45 degrees and made the green curve B vertical from the midpoint of that arc.
The blue curve is made using the Curve command and placing the 3 control points at the ends of curves B and C and the their intersection point

As I said before you could construct the Revolve and Filletsrf with history and then edit with the shape of the arching fillet by editing the shape of the blue curve. To keep it tangent with the vertical 10 mm fillet surf the 2 control points that are on curveC must stay on curveC. IOW they can be moved vertically to change the shape of the blue curve also the cv weights can be edited to change the shape. If you change the shape of blue then the red 5mm fillet (curveA) need to be made again to keep it tangent. to the two curves.

After seeing the picture of the actual bottle, this is what i think is roughly the shape of the blue curve.
bottle2.3dm (278.7 KB)

1 Like

thanks jim. now i get it!


thanks to everybody for your support. appreciate it!


Maybe rolling ball is a flawed concept. The concept makes “messes” when is ‘fails’.

Why not? Sounds like more than just a rolling ball problem.

Yet the curved face is attached to the same polysurface that’s attached to the selected edge…

Rhino should be able to trim all adjacent faces back and blend everything according to the selected adjacent edge, automatically.