Revision: 73850
http://sourceforge.net/p/brlcad/code/73850
Author: starseeker
Date: 2019-09-07 11:49:57 +0000 (Sat, 07 Sep 2019)
Log Message:
-----------
Need a somewhat different test here - right idea, but the CDT validity can't be
guaranteed at this point.
Modified Paths:
--------------
brlcad/trunk/src/libbrep/cdt.cpp
Modified: brlcad/trunk/src/libbrep/cdt.cpp
===================================================================
--- brlcad/trunk/src/libbrep/cdt.cpp 2019-09-06 20:00:35 UTC (rev 73849)
+++ brlcad/trunk/src/libbrep/cdt.cpp 2019-09-07 11:49:57 UTC (rev 73850)
@@ -1601,16 +1601,11 @@
// a valid triangulation for all faces. If not, refine the larger
// curved edges until we can.
//
- // In theory this is an expensive test, but as we are at the very
- // beginning of the process triangle counts should be relatively low
- // and all of these operations should be very fast. More critically,
- // in combination with the above this should give us the sparsest valid
- // triangulation we can return and provide a solid basis on which to
- // build.
+ // TODO - the below won't work in general, since without surface points
+ // we can't guarantee validity. Instead, check the expanded 2D RTree
bboxes
+ // to see if any of them interfere with each other. If they do, keep
splitting
+ // until the UV polygon curve bboxes are NOT self interfering.
//
- // Note that while the face will report individually valid, without
- // surface point sampling near the edges odd things may still happen in
- // the final mesh.
for (int index = 0; index < brep->m_F.Count(); index++) {
ON_BrepFace &face = s_cdt->brep->m_F[index];
cdt_mesh::cdt_mesh_t *fmesh = &s_cdt->fmeshes[face.m_face_index];
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
_______________________________________________
BRL-CAD Source Commits mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/brlcad-commits
