Tue, 09/08/2009 - 19:25
Forums:
Hello,
I´ve got some problems meshing a shape correctly. I´ve generated it with the BRepAlgoAPI_Common algorithm - maybe this alogrithm doesn´t produce correct results (but from the visualisation and from ShapeAnalysis_Shell it seems to be ok). I tried to mesh the shape with BRepMesh_FastDiscret and BRepMesh_IncrementalMesh using different parameters but the result is never satisfying. Especially with BRepMesh_IncrementalMesh I would have expected a different result.
Has someone an idea whats wrong - here the sample files.
http://rapidshare.com/files/277284509/MeshSample.brep.html
Thanks in advance - Stefan
Tue, 09/08/2009 - 23:35
I dont know if it make sense, but you could try to subdivide the original shape first with ShapeUpgrade_ShapeDivideArea?
Regards
Simon
Wed, 09/09/2009 - 12:59
Thanks for hint... but it has no effect on the shape. Maybe I´m using it in the wrong way!?!?
lShapeDivider.Init(lExplorer.Current());
lShapeDivider.SetMinTolerance(0.0001);
lShapeDivider.SetMaxTolerance(100);
lShapeDivider.Perform(Standard_True);
lResultingFace = TopoDS::Face(lShapeDivider.Result());
Wed, 09/09/2009 - 13:51
I think you have to set "ShapeUpgrade.ShapeDivideArea.MaxArea()= yourMaxArea" first, then call perform()...
Wed, 09/09/2009 - 16:05
Ok the tip is not bad.... but the "strange|wrong" discretising still appears at one side of the surface - so the problem is still visible. Has someone an idea whats wrong - I quess something is wrong with boundary or the edges but I have no experience with that!!??! :(
If wished I can upload screenshot or one more sampleFile!
Best regards - Stefan
Thu, 09/10/2009 - 13:37
Hmm I can´t understand why the meshing algoritm produces so irregular results. Here a screenshoot:
http://img245.imageshack.us/img245/9657/meshing.jpg
The first and second part have the same size - part three is a little bit different. Checked out the parameters but no one really helped - any suggestions what to try out.
Stefan
Thu, 09/10/2009 - 14:50
Perhaps http://www.hpfem.jku.at/netgen/ will produce better results? I dont know because I dont use neither the OCC mesher nor the netgen...
Good luck
Simon
Fri, 09/11/2009 - 16:12
Thanks for the tip Simon... but another Mesher is not an option for our team. :(
Fri, 09/11/2009 - 19:07
Ok after some investigations I understand whats the problem. The mesher seems to use the U|V - coordinates for its calculations. Because I´m using trimmed surface - these parametric values are mostly like u: 0.0-1.0 and v: 0.4-0.5. The small range of the v-values seem to make trouble.
Some ideas which could work - but I can´t implement them because of missing knowledge:
- trim a bsplinesurface for real
- create a new bplinesurface from a trimmed one (the new has parametric values between 0 and 1)
- approximate a trimmed face (GeomConvert_ApproxSurface doesn´t work for that)
Can one of these ideas be implemented???
Best regards - Stefan
Thu, 09/17/2009 - 08:41
maybe you could try to mesh the untrimmed surface first, and then trim the mesh within your trimmed surface boundary yourself. Again, the mesher seems to have a lot of bugs, as I heard.
Thu, 09/17/2009 - 11:34
Hmmm yeah... I quess this it will be - unfortunately. :(