Re: [help-3dldf] Fwd: Re: button-hole problem

[Laurence Finston]

> Take, for example, two parallel triangles.
> 
>                   A    B
>                   /\  /\
>                  /  \/  \
>                 /   /\   \
>                /  C/__\___\D
>               /________\
>              E          F
> 
> If AEF is in front of BCD, then
> 
>                   A    B
>                   /\  /\
>                  /  \/g \
>                 /    \   \
>                /      \h__\D
>               /________\
>              E          F
> 
> the plan containing A, F and Focus::position has cut BCD in two parts.
> Cgh is occluded and BghD is not.

Thanks for the explanation.  The case I'm having difficulty with is where the
triangles lie in a plane parallel to the x-z plane, the line Focus::position
-- Focus::direction also lies in a plane parallel to the x-z plane, and the
triangles as seen from above look something like this, but rotated about  plus
or minus 75°:

______
|           /|
|          / | 
|         /  |
|  A   /   |
|      / B |
|     /      |
|    /       |
I   /        |      
|  /         |
|/_____|

I've implemented routines for "self-decomposition" of polygons, but I'm afraid
that no matter how many pieces I break them into, simply sorting by their
max-z, min-z, or mean-z values will fail for some of them.  I may try
projecting a ray from Focus::position through each of the vertices and testing
the distance from Focus::position to the first intersection point of each of
the polygons.  I think this will work for convex polygons.
I don't know what to do about conic sections yet.

Thanks.

Laurence