So I am fighting with the Sweep SOP, again. Despite Peters tutorial there are still some matters not perfectly clear to me.

It seems the only way to get a decent result from the sweep SOP is that when the section and the sweep curve are planar and perpendicular to each other, the section profile is created on the XY plane and at the origin. Isn't this a huge limitation of using the sweep SOP in any procedural network?

Attached a file with four polygon surface and a polygon curve. The results on the differ quite a lot. Why? How can I make this setup in such a way that I can throw in any curve and get the result as on the first polygon surface.

Thanks,
Hans

It doesn't exactly answer your question, but why not just sweep the base curve then copy the result 4 times? Less work and less nodes.

Bah, i am not satisfied with that answer at all Graham, too easy.

Okay, so this was just a little quick dirty setup to show the different results of a sweep on a rotated curve. I am working on an otl where i want to sweep a profile along a curve, so the curve could have any position, size and orientation in the scene. How can i make the sweep fool proof so that I get the result as on the first sweep curve?

With the current way sweep works (at least, as it seems to me), i have to transform the curves in the otl to 0,0,0, then rotate them in such a way they are perpendicular to the cross-section, execute the sweep and move things back to their original position.

Hans