Why is this branch so painful?

I’ve thrown everything I can think of on this one. From simple 4curve networkSrf to elaborate patches, sweeps, split edges…
It will not produce a smooth transition.
PainBranch.3dm (125.3 KB)

Edit:
This is the closest I got (after posting the question), but it involves blending the two corners from scratch instead of using the existing guide curves. (those guide-curves were blends themselves but were later projected to an underlying surface which has a curvature of its own; not shown in these example files)
Possible Solution.3dm (69.7 KB)

Hi Thomas,
this doesn’t answer your question, but this is a typical scenario where SubD excels over Nurbs.
This is done in V6 just for fun with some custom tools.
I just made a mesh based on your surface’s controlpoints, but it isn’t perfect regarding pull-force (I have a tool that can make a mesh from scratch, or continue on one, by picking 3 or 4 corners for the new face)

1 Like

Yeah, these are tricky, at least in part because you’re asking the transition to match a surface both along and across its principal curvatures - and in between. Just for the sake of variety, this construction might be a way to think about it

PainBranch__PG.3dm (80.9 KB)

Also. I adjusted the lower ‘incoming’ surface just a bit as it seemed to be ‘pointing’ such that it misses the larger surface, so to speak-

I made that red line come in a little more onto the other surface.

-Pascal

1 Like

Good one. I wonder if it would be feasible to apply the SubD type of relaxation technique to a Nurbs tool (something that preserves the edges).

ok, first off (giving credit were it is due) it requires God level of mastery to create such thin objects (a hair from being called out as “bad objects”) and still have a flawless mesh and continuity !

This method is a “main centric” split where a large branch becomes two distinct small ones with a distinct crease in between (like a piping system where fluid would split two ways from a main). I was more after an equal three way blending where each of the three branches blend in both the other two directions.

… but I really like the end result of this the more I inspect it. Thanks for taking a hack at it.

Hi Thomas - I’d probably not let that one surface peter out to a point - I’d give it a little width at the pointy end if I were to do it again. Here’s a more ‘equal’ layout, I tihnk you had something similar:

PainBranch__PG_MoreEqual.3dm (1.2 MB)

-Pascal

Yup, that’s exactly it. I think you are using a two rail sweep at the top (where I used a patch instead). Thank you.
(by the way you haven’t skipped a beat in over 15 years, as sharp as ever)

Hi Thomas - just to be picky - it was not a Sweep2 but could have been, or a BlendSrf… I used a loft between sections of extracted isocurves and then MatchSrf to the ‘parent’ surfaces. I did end up using InsertKnot to get a few more points in there to clean up the surface match. The thinking is this process gives me more control over how dense the surface gets - I try only to add points if I need them, it just keeps things more fair and editable.

-Pascal