Limitations of Edge softening (?)

I am noticing objects like these generally fail to soften.
Objects like these also fail to fillet. (Is this the reason softening also fails even though it’s mesh based?)
EdgeSoften.3dm (171.7 KB)

Hi Thomas - there is a small sliver of surface here:

If you MergeAllCoplanarFaces that cleans it up and the softening works (albeit not all that well in this case, I’ll play with the mesh settings)


Wow, good catch on that little sliver. How did I miss it.
Though that was an error introduced while doing this example quickly to illustrate the question on softening. I have geometry that is clean in the past and had to manually fillet the whole thing.

Not complaining, or asking you guys to change anything. Fillets are very complex topic and I would prefer McNeel iron out other simpler things first before these kind of thing.

Basically the question is. If the topology is such that the main NURBS fillets fail we should also expect the softening ones to fail ?

Hi Thomas - I reconstructed the thing with simpler, presumably ‘better’ surfaces and I now get the error you got if I go much above .1 on the softening - so I am still poking at this.
My mistake - I capped the object not seeing there was a skinny opening and created another skinny surface in the process.

Your surfaces are fine - I just went as simple as possible to see if I could make the edge softening work more cleanly - medium success-

EdgeSoften_PG.3dm (168.3 KB)

This is a different ball of wax from actual filleting, it is a display mesh trick. But edge filleting would also fail at that location; with surface filleting you could skip over the skinny surface and get the larger plane though, and that would work.


Yes, it’s very possible I am stuck with old surfacing tricks. If it wasn’t just an example I would likely guide those networkSrf with a blendCrv (from Isoparms at the middle) for a smoother transition.

I have to say, I’ve been impressed in the past by your years of surfacing refinements tempered through thousands of interactions with users.

ok, I see what you did.
The reason I avoid surfaceFrom 2,3,4 edge curves in these cases, is that on a real project you’d want to control the curvature transition in the middle via the red guide curve.

I see the “soften” is trying to do something.

@pascal I am noticing sub-D maybe the way forward with these types of organic hinges.
Thanks to @BrianJ for his online SubD tutorials, getting us started.
SubD.3dm (144.0 KB)

If you were to do this with SubD, how do you cut two crescents from the base of the top pancake ?
SubD.3dm (95.8 KB)

1 Like

Hi @ThomasAn,

I’d actually start with Pascal’s clean version in NURBS and then QuadRemesh it into a SubD… super fast and with X and Y symmetry looks pretty good in this case. More faces than strictly needed but I’d take it for the ease of use. If I needed a tighter fit to something I would then Pull the SubD control points or verts to another surface.

Although splitting a SubD not along existing edges and keeping it a SubD is a filed feature request on our end, you currently have to play by the edges and delete faces to make a hole.

1 Like

A hybrid approach :thinking: … I take it would be more fiddly doing this in subD from scratch. I spent about an hour on a “from scratch” version and was getting caught in edge loop routing decisions and Ngon fixing. Concluded it be longer doing it all subD from scratch.

I love hybrid workflows :slight_smile: don’t even get me started on voxels


Does Rhino have the equivalent of Blender’s spherical command ? In other words, is there a way for the points of the top square to be made round ?

( Blender Secrets - Make selections Circular with To Sphere, Circle and Smooth tool - YouTube )

No… but I filed it , I’ll add another request thanks. Right now you have to use the Align command with the ToCurve option to put control points or verts on an actual circle.

1 Like