Surfacing/Filleting Problem - 3 way corner



Hi folks,
I’m having trouble with a blend surface around a three way corner. (I don’t do this sort of blend very often).
The three fillets are different in size and this is creating some edges that are creating problems. Can anyone advise on how to make this blend smoother? I’ve uploaded the file below. Fillet is in blue.

3 way corner blend problem.3dm (176.4 KB)

Michael Deimen

(Diego Krause) #2

Since the difference between radius is too big, you can try with two surfaces instead of one


Hi Diego,
Hadn’t thought of that. I’ll give it a try. Thanks for the help.


Hi Diego,
Your solution worked really well for the corner I sent you first. Then I tried it on a second corner farther around the part. (See attached.)Didn’t work as well there. Any thoughts on how to dial that one in (again, in blue).

3 way corner blend problem_2_.3dm (105.5 KB)

Thanks again,

(Pascal Golay) #5

Hi Michael - it looks like the same basic idea will work pretty well -

3 way corner blend problem_2_PG_.3dm (141.6 KB)



Hi Pascal,
I’m seeing a bit of a crease between surfaces A and B (see image below). Are you not seeing this?

Michael Deimen

(Pascal Golay) #7

Hi Michael - - Zebra looks decently tangent there -

What if you Join the surfaces, does that look cleaner?



Hi Pascal, All three surfaces in question were joined in the file I uploaded. Here’s the Zebra analysis…


Apologies Pascal. I thought all the surfaces were joined, but the long surface does not seem to want to join to the first blend surface. Any thoughts on what to do with that?

Thanks for the help,


This is really odd. In the original file the surfaces join, but in the extracted file they refuse to. Here’s the zebra from the original file where the surfaces are joined. Still getting a bit of crease there.


(Pascal Golay) #11

Hi Michael - you’re right, sorry - I was looking at the wrong edge - InsertKnot >Automatic, twice on that surface to give it a bit more control - you’ll see a couple of extra isocurves appear ,

then MatchSrf for tangency to the blue fillet surface once more.

@Michael_Deimen - another thought on the first case - for what it is worth - this way maintains the straight fillet as far as possible then makes the transition on the arc part of the corner only…

3 way corner blend problem_2_PG2_.3dm (146.4 KB)



Since the fillets you are trying to connect to all appear to have been round to begin with why not use a standard round fillet solution? Trying to mix round fillets with unround blends is a lot harder than just using all blends or all round fillets.

a3way_corner.3dm (90.5 KB)


Hi Jim,
I couldn’t seem to get the standard fillet approach to work. It either failed altogether or looked awful. That said, its entirely possible I was doing something wrong.



Hi Pascal,
I did the insert knot twice as advised. Not sure I did the MatchSrf properly. It took, but still getting a couple creases at the edges of this surface. I built it as a network surf. Would a patch work better? Just a thought.


(Pascal Golay) #15

Hi Michael - what does Zebra tell you there? Are they actual creases? Here’s what I see (matchSrf both edges for Tangency)



Hi Pascal,
It may be that I need more education in how to read the Zebra analysis, I would have read this as a break in surface continuity. I’m assuming from your post that as long as the stripes connect at the same points and same width without break, I have good continuity.
I actually went back and rebuilt the surfaces using “ArcBlend” at the edges and “Adjustable Curve blend” to add more curves in the middle of the surface. Those I set at “Tangent” on both ends, although I experimented with “Curvature” and “G3”. The ArcBlend curves lifted the middle of the edge curves a bit so they were in closer alignment before I built the other network curves. In the end the surface I have look like this on Zebra Analysis:
Lower Corner

Upper Corner

Not sure I can get this any better without a refresher in Blending and Filleting.

Your thoughts?



Tangency to the left (“stripes connect”) and Curvature to the right :

Edit: Better picture:

// Rolf

(Pascal Golay) #18

Hi Michael - one thing is, when using false color display like Zebra and CurvatureAnalysis, make sure to boost the analysis mesh to a very fine setting (Adjust Mesh button) - the cheap and easy way to do this is to set the detailed controls > Minimum initial grid quads to a large number - 3,000, 5,000, even 10,000 and leave the other controls zeroed out. Then you can get a good read and not be distracted by a chunky analysis mesh messing up your stripes.



Hi Pascal,
This is the geometry I have now. Been working on it all day. I think its close but I doubt its perfect. Heading home for the day now, but I’d love to hear any ideas you may have for improving it.

3 way corner blend problem_3_.3dm (116.2 KB)


(Pascal Golay) #20

Hi Michael - it looks like you are shooting for curvature continuity - this may or may not be practical everywhere here - this is a filleted thing, basically, i.e. the existing surfaces are tangent and not G2, so in general you’ll want to make new surfaces that are also tangent to one another, at least where they run up against two existing surfaces that are tangent.

I’d still go after this with simpler surfaces though.

3 way corner blend problem_2_PG2_V5_.3dm (156.0 KB)