[Paraview] Extents in PVTI Files
Berk Geveci
berk.geveci at kitware.com
Fri May 9 11:50:03 EDT 2008
Oops. I didn't read the first message carefully. Yes, the points at
the boundaries must overlap.
-berk
On Fri, May 9, 2008 at 11:45 AM, Kent Eschenberg <eschenbe at psc.edu> wrote:
> Thanks! That made no difference. The first vti file uses
>
> WholeExtent="0 127 0 127 0 127"
> Piece Extent="0 127 0 127 0 127"
>
> and the second one now uses
>
> WholeExtent="128 255 0 127 0 127"
> Piece Extent="128 255 0 127 0 127"
>
> Using 32 proccesses the error messages say the following extents cannot be
> filled:
>
> 127 128 0 63 0 63
> 127 128 0 63 63 95
> 127 128 0 63 95 127
> 127 128 63 95 0 63
> 127 128 95 127 0 63
> 127 128 63 127 63 95
> 127 128 63 127 95 127
>
> This may be irrelevant but its interesting to note that all include the
> "gap" at (127,128) between the two vti files. So I again ask, is there an
> unstated requirement that the vti files must overlap?
>
> TIA!
> Kent
> Pittsburgh Supercomputing Center
>
>
> Berk Geveci wrote:
>>
>> Kent,
>>
>> I believe the problem lies in the extent of the vti files. They cannot
>> be 0 127 0 127 0 127. They have to be with respect to the "global"
>> extent. For example, one of them may be 128 255 0 127 0 127...
>>
>> -berk
>>
>>
>> On Thu, May 8, 2008 at 5:27 PM, Kent Eschenberg <eschenbe at psc.edu> wrote:
>>>
>>> I'd like to use a pvti file to read many vti files but the attached
>>> example
>>> fails with messages like this:
>>>
>>> ===
>>> Process id: 9 >> ERROR: In
>>>
>>> /usr/local/packages/paraview/Edge/ParaView3/VTK/IO/vtkXMLPStructuredDataReader.cxx,
>>> line 452
>>> vtkXMLPImageDataReader (0x1a41c8f0): No available piece provides data for
>>> the following extents:
>>> 127 191 63 127 0 63
>>> The UpdateExtent cannot be filled.
>>> ===
>>>
>>> Each of the vti files can be displayed just fine and have these
>>> parameters:
>>> origin 0 0 0
>>> spacing 1 1 1
>>> extent 0 127 0 127 0 127
>>>
>>> In the pvti file I'm a little confused about the definition of extent and
>>> cannot find the answer in the VTK file format guide. The key lines in my
>>> file are these:
>>>
>>> <Piece Extent="000 127 000 127 000 127" Source="test0.vti"/>
>>> <Piece Extent="128 255 000 127 000 127" Source="test1.vti"/>
>>>
>>> I've seen examples where the two files overlap by one point so that the
>>> above two lines would read
>>>
>>> <Piece Extent="000 127 000 127 000 127" Source="test0.vti"/>
>>> <Piece Extent="127 255 000 127 000 127" Source="test1.vti"/>
>>>
>>> Is that an unstated requirement for pvti? I'm using ParaView from CVS
>>> around
>>> 5/6. This problem occurs when running standalone or with a server running
>>> in
>>> parallel on a Linux cluster.
>>>
>>> Kent
>>> Pittsburgh Supercomputing Center
>
More information about the ParaView
mailing list