Well, I’m not sure why the blend tool isn’t giving you a nice result, it should be fine if you have the settins right, but I think you need to back up a bit…what’s your pan for making the actual top “surface?” I don’t see this working.
I think the latter is better, since the final goal is always a surface (not curves) and the faster you can get there the earlier you can detect visual flaws/problems with the surfaces. I try to “think” in surfaces when I model, not in a sequence of edges form which surfaces hang of. – if it makes sense
So after some time I have another question concerning your second method.
You say to construct the curve as above. After trimming with the circle, do I also blend the curves?
and after I constructed the outer curve ring, what is the best way to get the inner?
Offsetting leaves me with with way too many control points
Does the inner curve need to be an “exact” offset of the outer curve? If so the large number of control points are necessary because of how NURBS math works. Except for special cases the offset of a NURBS curve cannot be represented exactly by a NURBS curve. Rhino uses enough control points for the offset curve so that it is within the tolerance of the exact offset.
If the Loose=Yes option is selected in Offset then the offset curve will have the same number of control points as the original curve but may deviate from the exact offset.
hm, It was a bit unclear: I would not construct the blend curves( “corners”) – only the sides. Trim with a circle around the corner; then extrude curve ( or extrude curve tapered ). After this you can “blendSrf” the corners. This is not mathematical perfect (the inner curve is not an exact offset) but it is faster and you have a cleaner control point layout. In rhino V7 the blendsrf command is even better an doesn’t add unnecessary control points.