Hi Pilou - in your case each piece of your "sine wave" like curve is a full 180 degree arc.

That's a lot more exaggerated of a bulge than a normal sine wave curve.

If you make the bulge of each of your pieces less extreme you will likely get a less bulgy result...

With that large amount of bulging, it would produce a very complex sweep as it tries to rotate the starting shape along each of those exaggerated bulges... That's why some kind of non-rotating method is needed.

You can kind of see what you are asking it to do if you do Transform/Array/Curve on it to make some copies along the curve that are rotated, you'll see a result like this:






You can kind of get an idea there of how your profile curve will swing around as it is rotated along the path, then it tries to produce a surface through that...

- Michael

Swinging indeed like smoke simulation! :)

i was just playing with this for fun. you can truly sweep 1 segment and then array it.

edit: i think i might have done that with 'flat' sweep anyway. d'oh. :)

Yes it's the trick as show above ;)

Works pretty good with the extrude-path. Here is sin wave "amp=4 length=20 cycle=10 points=5"

Sorry. I just noticed your talkin about your "particular" curves. My bad.

Ack, this one broke flat sweep. Extrude path worked. Better stop now. :)

In fact seems nature of curves, & orientation are very sensitive!
this one works fine with Network! (5 seconds on a very old comput)
A small change and all works fine or not!
So a big training is necessary :)