I was testing an approach suggested here for shaft bossing but have run into a bit of a mystery.
I am trying to use BLENDSRF. At the top there is no blending other than by position. At the bottom I do get a curve.
Not sure yet why it does this - there are discontinuities between the inputs… still poking but here’s a workaround that may help - the white surfaces are blends the green is a Sweep2.
@miano - shrininking this surface (ShrinkTrimmedSrf) and then SetPt on the control point right at the corner to stick it exactly on the corner of the surface below, sorts it out. I would agree this should not be needed - it looks like the not quite coincident end points of the input edges is forcing the blend points to be almost on top of each other along the edges.
the coincidence is the issue, works fine if you trim the surface mentioned by Pascal.
I would personally not end a blend surface with a coincidence either, I would rather make is turn at the end and create the other side in one blend by using a C2 bend curve and cut between the two outer edges
So far I have not been able accomplish anything from coincidence. On my end they already look next to each other, except that intersecting the two surfaces gives nothing. However, dupedge and intersect does show coincidence
Your half cylinder in the middle should ideally be in curvature continuity with the two straight side wall, by either creating a G2 curve between them, or approximating the circle with a single spline in G2 between the straight edges, you can then do a dual rail which would give you a single surface vs polysurface made of 3 surfaces not in continuity, much cleaner outcome on the blend