> Your curved walls are presently cut by boxes or planes that are oriented
> radialy. OK for self standing walls or curved walls that need to
> continue as straight ones.
> They should be cut by box or plane that are parallel to the wall the
> curved wall connect to, then you need to also remove the rest of the
> cylinder.

Yep! That seems to be the only interesting solution right now.

> There is also nothing wrong in over shooting and hiding part
> of the curve inside adjacent walls. This should not cause any problem as
> long as the curved wall is the same thickness, or thinner, as the other
> walls.

That will not be acceptable if the intention is to actually use the resulting
model to construct a building which is the goal of kobldes.

> Using prisms, you need to generate many points that are placed along a
> circle. Maybe as many as 1 point for each degree along the curve. Then,
> you can use a cubic_spline to smooth out the curve.

I experimented with this for quite a while but the curve produced was not
consistent. It may be perfectly curved for a couple of circumstanced but there
was always a situation when it would throw a dark spot or a misaligned edge
somewhere. Generally handling many UV values made it too tedious for me. It was
no longer fun!

> You can use sin(angle) and cos(angle) and multiply those by the actual
> radius of the wall. It's then easy to have different start and end angle
> for the inner and outer surfaces. Just make sure that you use a slightly
> longer arc to hide it's ends inside the other walls.
>

Perhaps an illustration? I think chopping up the ends of cylinders as you
alluded to in the beginning is a much more interesting solution.

Post a reply to this message