How do I avoid the wrinkles created by BlendSrf in this model

In the attached model I have used BlendSrf to join two parts, the larger of which is made from multiple surfaces joined (so I use the ChainEdges option). In the blend I get two areas of highly concentrated isocurves that appear to represent small ripples in the blend surface, as though it is made up of sections that are slightly misaligned. These originate where chained edges of the larger part meet. I believe that the joins in the larger part are clean so I don’t see an obvious reason for these discontinuities in the blend.

So is this a bug or known behaviour that wiser heads than mine can explain?

And what can I do to get rid of them (or better, avoid them)? I’ve tried trimming out the concentrations in the blend surface at the adjacent isocurves and blending the resulting edges but I’m still left with imperfections.

MergeSurfacesWrinkle.3dm (405.4 KB)

This is from Rhino 6 SR7


Your trim curve has way more control than it needs.

Here I’ve extracted the edge of the polysurface and turned on points:

I’d untrim the surfaces, cleanup and simplify the trim curve, and retrim the surfaces.
My guess is your tolerance of 0.0001 cm is the cause.
Since you’re modeling a large object in cm, why such a tight tolerance?

Hi Jeremy - this looks hard to get done cleanly as is … the middle and lower surfaces are tangent, not curvature continous (btw the middle and right surfaces are not tangent, probably should be - but there is not such a big ‘glitch’ there, I suppose because the curvature is more parallel there where as on the problem area the trim is more at an angle to the surface curvature) So the upshot is - ideally, match the three input surfaces for curvature, then trim that crescent out and the blend should look cleaner.


Hi John, Pascal,

I wanted to mark both your replies as the solution as both sets of advice were needed, but discourse only allows one selection…

The trim line is from a simple ellipse in the YZ plane so the high number of control points is generated in the trimming. Following John’s suggestion, I reduced the model tolerance (to 0.01) and retrimmed, getting a lower point count. The subsequent G3 blend went haywire. But when I rematched the compound surface edges for curvature instead of tangency prior to retrimming, per Pascal, then the G3 blend worked well and I lost the wrinkles. Note that if I apply the curvature matches but not the tolerance change the wrinkles still appear.

John, I can’t remember why the tolerance was set so tight - this is part of a self-inflicted exercise involving a much more complex model that I’ve been doing off and on over the past month or so to improve my surface modelling skills (by instinct I’m a solid modeller) - and I am pretty sure it was done to make something else work, but it might simply be how the original model I inherited was set up.

Thanks for double-teaming this one guys!