[Paraview] paraview crashes on obj file originating from VMD

[Peter C. Lai]

Thanks for the reply. Any conversion alternatives? Like converting from
POVRay? 

Let me look into inlineing the VRML mats. I will probably panic more
about this in a month when stuff needs to work, but thanks for expressing
and interest.

On 2011-02-17 12:19:20PM -0600, David E DeMarle wrote:
> For VRML it looks like our ancient reader does not support aliasing.
> If you can configure the code that created vmd1.wrl to inline the
> VMDCyl and VMDMat, I suspect that ParaView will read it.
> 
> For our equally ancient OBJ it does appear that we do not support mtl
> files. The vtkOBJReader.cxx has a comment (following) that describes
> the small subset of OBJ that it will read. I was able to read the file
> with 3.10 with a debug build, but not a release build, which indicates
> an uninitialized variable in the reader, which I should be able fix
> for 3.10 final.

Hmmm, I'd love to see the patch for this (I tend to build it from source
to begin with).

Also, if it reads texture coordinates, doesn't mtl make up part of the texture
definition? (and mtl is basically just a list of RGB triplets for
ambient, diffuse, and specular colors, and a single scalar each for alpha chan, 
phong, and illumination model, with an option for including a texture map);
I wonder if it's just a stub function we can replace (since the renderer
has to apply the default white texture somehow anyway...).

> 
> Please suggest modernizing these readers on user voice and if enough
> people chime in we'll get it into 3.10.1 or 4.10.
>

User voice?
 
> 
> /*---------------------------------------------------------------------------*\
> 
> This is only partial support for the OBJ format, which is quite complicated.
> To find a full specification, search the net for "OBJ format", eg.:
> 
>     http://en.wikipedia.org/wiki/Obj
>     http://netghost.narod.ru/gff/graphics/summary/waveobj.htm
> 
> We support the following types:
> 
> v <x> <y> <z>
> 
>     vertex
> 
> vn <x> <y> <z>
> 
>     vertex normal
> 
> vt <x> <y>
> 
>     texture coordinate
> 
> f <v_a> <v_b> <v_c> ...
> 
>     polygonal face linking vertices v_a, v_b, v_c, etc. which
>     are 1-based indices into the vertex list
> 
> f <v_a>/<t_a> <v_b>/<t_b> ...
> 
>     polygonal face as above, but with texture coordinates for
>     each vertex. t_a etc. are 1-based indices into the texture
>     coordinates list (from the vt lines)
> 
> f <v_a>/<t_a>/<n_a> <v_b>/<t_b>/<n_b> ...
> 
>     polygonal face as above, with a normal at each vertex, as a
>     1-based index into the normals list (from the vn lines)
> 
> f <v_a>//<n_a> <v_b>//<n_b> ...
> 
>     polygonal face as above but without texture coordinates.
> 
>     Per-face tcoords and normals are supported by duplicating
>     the vertices on each face as necessary.
> 
> l <v_a> <v_b> ...
> 
>     lines linking vertices v_a, v_b, etc. which are 1-based
>     indices into the vertex list
> 
> p <v_a> <v_b> ...
> 
>     points located at the vertices v_a, v_b, etc. which are 1-based
>     indices into the vertex list
> 
> \*---------------------------------------------------------------------------*/
> 
> 
> 
> David E DeMarle
> Kitware, Inc.
> R&D Engineer
> 28 Corporate Drive
> Clifton Park, NY 12065-8662
> Phone: 518-371-3971 x109
> 
> 
> 
> On Fri, Feb 4, 2011 at 6:42 PM, Peter C. Lai <pcl at uab.edu> wrote:
> > Hello
> >
> > We need to export molecular structures and surface renders from UIUC's VMD
> > to OBJ with color/material support into Paraview for use with VisBox.
> > Paraview 3.8.1 and 3.10-RC1 both crash while trying to render the OBJ files
> > here:
> > http://cesium.hyperfine.info/~cowbert/vmd/vmd1.obj
> > http://cesium.hyperfine.info/~cowbert/vmd/vmd1.mtl
> >
> > Segfaults after hitting 'Apply'.
> >
> > Furthermore I read somewhere that the vtk OBJ reader does not read or process
> > MTL files? Is this something that can be fixed? We'd really like surface
> > colors and transparency to be imported from the mtl file...
> >
> > Also Paraview will also crash reading in VMD output to VRML2 with a segfault:
> > http://cesium.hyperfine.info/~cowbert/vmd/vmd1.wrl
> >
> > The VMD developers at UIUC are certain these are problems with Paraview and
> > not their software...
> >
> > One final method might supportable is: output POV-Ray from VMD and then
> > convert it to something that Paraview will read. Any ideas on that?
> >
> > Thanks
> > --
> > ===============================================================
> > Peter C. Lai                 | University of Alabama-Birmingham
> > Programmer/Analyst           | BEC 257
> > Genetics, Div. of Research   | 1150 10th Avenue South
> > pcl at uab.edu                  | Birmingham AL 35294-4461
> > (205) 690-0808               |
> > ===============================================================
> >
> > _______________________________________________
> > Powered by www.kitware.com
> >
> > Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
> >
> > Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
> >
> > Follow this link to subscribe/unsubscribe:
> > http://www.paraview.org/mailman/listinfo/paraview
> >

-- 
===============================================================
Peter C. Lai                 | University of Alabama-Birmingham
Programmer/Analyst           | BEC 257
Genetics, Div. of Research   | 1150 10th Avenue South
pcl at uab.edu                  | Birmingham AL 35294-4461
(205) 690-0808               |
===============================================================