Le 11/02/2014 14:47, Robert McGregor a écrit :
> The starting position for v shouldn't matter as it's a seamless loop around the
> circumference (like cylindrical mapping).

it's not the starting position, it's keeping it coherent along the
various components... maybe a target position such as <0,0,0> to define
the 0, and a clockwise or anticlockwise rotation to increase the value.

Yet it fails when the spheresweep pass through the target position.

Putting the target position at some infinity does not work either, when
the direction of the cylinder/conic is aligned with that direction.

An easy cheating would be u only, v is always 0 (or 0.5), but it's bad.
You can ignore v, but it should be defined... and not constant.

It's also problematic when on a connecting sphere (at one of the vertex).

-- 
Just because nobody complains does not mean all parachutes are perfect.

Post a reply to this message