From ba34ff418c829c64fd7c39ee6484b9e172bb9854 Mon Sep 17 00:00:00 2001
From: mattijs <mattijs>
Date: Fri, 17 Jun 2011 17:24:56 +0100
Subject: [PATCH] ENH: differencesWrtDev.txt:update

---
 differencesWrtDev.txt | 40 +++++++++++++++++++++++++++++++++++++++-
 1 file changed, 39 insertions(+), 1 deletion(-)

diff --git a/differencesWrtDev.txt b/differencesWrtDev.txt
index 3d87e6d01a..f3b0568ca2 100644
--- a/differencesWrtDev.txt
+++ b/differencesWrtDev.txt
@@ -1,7 +1,7 @@
 src/
 
 mesh/conformalVoronoiMesh
-    - all the meshing.
+    - all the meshing. See separate section below.
 
 polyMeshGeometry:
     - disabled tetquality check on face-center decomp tet.
@@ -93,5 +93,43 @@ surfaceSplitByTopology:
 
 
 
+conformalVoronoiMesh/
+--------------------
+- Make sure the surface does not have any sliver triangles. These are
+hard to get the surface normal correct for so might cause bleeding.
+- Use surfaceCheck to find out triangle quality and size of smallest edges
+- Use surfaceClean .. 1e-5 .. to get rid of any edges < 1e-5m.
+
+- If you get bleeding you might see in face filtering:
+
+    ...
+    cells with with zero or one non-boundary face          : 1
+    ...
+    Initial check before face collapse, found 48 bad quality faces
+
+but this was real - the cell that got created inside the cone and sphere by the 
+bad triangle was actually attached to a lot of faces.  This screwed up the 
+subsequent filtering as it stopped too early.
+
+I ran:
+
+    surfaceClean coneAndSphere.obj 1e-5 coneAndSphere_clean.obj
+
+and re-ran with that surface and got
+
+    ...
+    cells with with zero or one non-boundary face          : 0
+    ...
+    Initial check before face collapse, found 0 bad quality faces
+
+and the bad cells inside are gone.
+
+That group of cells would be picked up at the end of the meshing as the
+
+    cvMesh_remainingProtrusions
+
+cellSet which can be deleted.
+
+