[vtk-developers] Fwd: Surface

Mark Roden mmroden at gmail.com
Mon Feb 21 11:40:31 EST 2011


Hi VTK developers,

There appears to be some problems running decimation routines of large
volumes in parallel.  This is a bug report forwarded by one of the
gdcm contributors for their large project involving both gdcm and vtk.

Thanks,
Mark


---------- Forwarded message ----------
From: totonixsame at gmail.com <totonixsame at gmail.com>
Date: 21 February 2011 10:21
Subject: Superficie
To: Paulo Henrique Junqueira Amorim <paulojamorim at gmail.com>


Hi,

In InVesalius we generate a vtkPolydata from a medical vtkImageData
using marching cubes. But the vtkImagedata we are working are very
large, like 512x512x1000, in a 32-bit OS it crashes and is a lot slow.
So we thought in parallelize this part of software, then InVesalius
could generate vtkPolydata faster. To do that firstly we split the
vtkImagedata in smaller pieces, to each pieces we do this pipeline:

vtkMarchingCubes -> vtkTriangleFilter -> vtkDecimatePro
(PreserveTopologyOn and SetTargetReduction(0.3)

Each piece can run in parallel.

After all piece are processed we join them, using this pipeline:

vtkAppendPolyData -> vtkCleanPolyData -> vtkStripper

Then it shows to the user. But it has a problems: In junctions there
is some marks like showed in [1] and [2]. When I open the STL
generated from that vtkPolydata in Meshlab, for example, in those
junctions there are some holes [3] and [4].

Do you have some idea why that problem occurs? I have to use some
other filter(s)?

Thanks!

[1] - http://i.imgur.com/MLXPw.png
[2] - http://i.imgur.com/DRFHy.png
[3] - http://i.imgur.com/0RB45.png
[4] - http://i.imgur.com/70fIa.png



More information about the vtk-developers mailing list