Hi Paolo, the part that you don't like there is the blending that happens as one profile is being morphed into the shape of the next or previous profile in the sweep.
This morphing is done using a kind of "ease-in / ease-out" method which results in the shapes that you see there.
That style of blending allows for the surface to be smooth at that center point.
If there was no kind of ease-in / ease-out at all and just a direct averaging kind of blend, it would result in a shape like this:
That does not work very well because the surface has an abrupt shift in shape where it hits a profile curve.
The current blending works with this kind of a function to make a smooth step from one profile to another (in the case of a pointy end then one of your profiles being blended is a point):
It sounds like you want a blending profile more like this instead:
Unfortunately it is not an easy thing for me to just switch everything over to use a blending mechanism like that instead - one mechanism is not automatically "better" than another and actually in other cases the current blending function is probably the more expected result, for example here:
If the blending worked the way you are talking about, the tube would have kind of "fat ends" where the center profile would have a sort of more exaggerated weight than the ending profiles.
The current mechanism treats each profile with equal weight which I think is probably the best for a default processing.
So it is not really an easy thing for me to change without causing undesirable side effects.
For now at least if you want to have more exact control over the way the sweep is shaped, a 2 rail sweep will be a better choice for that like you showed previously since that gives you a different method to control how the profiles are shaped since they will continuously stretch to match those guide rails in that case.
With one-rail sweep (without a scaling rail) the only method applied to the profiles is the ease-in / ease-out blending and nothing else affects them.
- Michael
