RE: [Algorithms] LSS-triangle test

[Bob D. <Bob...@bl...>]

> Yeah, you're right, the sphere doesn't rule out the edges,=20

Hmm... Charles, what you're saying sounds (to me :)) like it might work. =
Our current approach is to check (look ahead) which edge will be hit =
first then check that edge (along with it's endpoint). The idea being =
that if you're slightly (numerically) out and select the wrong edge then =
the correct answer is probably the vert (and this should come out in the =
wash). But your method of checking infinite cyls/segs sounds like it =
might yield (or be persuaded to yield) an essentially equivalent set of =
results with advantage in the intermediate/early-out dept. I'll take a =
closer look when I'm next in that area of our codebase. :)

On the accute/obtuse question... An interesting exercise is to draw a =
collection of (three say) radius-expanded tri corners with Voronoi =
regions for the two edges and associated verts. Look at the case of a =
small angle (<pi/2), a right-angle and a large angle (pi/2<theta<pi). =
For small angle the "vertex region" (in the outcode sense) is smaller =
than the Voronoi region for the vert; for the right angle case the =
Voronoi region and "vert region" are the same; for the large angle... =
etc. I think here lies the answer as to how the different cases, erhh... =
differ.

:)

Bob.


> -----Original Message-----
> From: Charles Bloom [mailto:cb...@cb...]
> Sent: 06 August 2004 18:39
> To: gda...@li...
> Subject: RE: [Algorithms] LSS-triangle test
>=20
>=20
>=20
> Yeah, you're right, the sphere doesn't rule out the edges,=20
> they may be=20
> sooner, but a hit on the front side of a finite cylinder does=20
> rule out the=20
> sphere and the other cylinder.
>=20
> At 12:09 PM 8/6/2004 -0500, Jeremiah Zanin wrote:
> >Are you sure you can ignore the edges if you hit the vertex=20
> sphere in a 2
> >edge region?  I think you can hit an edge closer than the=20
> vertex sphere so
> >you have to check the edges also.  If the line segment is=20
> clipped by an edge
> >then you wouldn't have to check the sphere though, if it's=20
> not clipped by an
> >edge then you have to check the sphere...I think.
>=20
> --------------------------------------------------------------
> ------------------------------
> Charles Bloom    email "cb2"    http://www.cbloom.com=20
>=20
>=20
>=20
> -------------------------------------------------------
> This SF.Net email is sponsored by OSTG. Have you noticed the=20
> changes on
> Linux.com, ITManagersJournal and NewsForge in the past few weeks? Now,
> one more big change to announce. We are now OSTG- Open Source=20
> Technology
> Group. Come see the changes on the new OSTG site. www.ostg.com
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_id=3D6188
>=20