[Paraview] Comparing datasets - taking a difference

[amayt]

Thanks a lot!

One more question.  If one can create difference maps, they would also
like to know statistical details of them (e.g., mean, standard
deviation).  I know the histogram filter is useful for this, and data
ranges can be found in the information tab.  However, there seems to
be no ways to know more details.  Actually I'd like to calculate the
mean as well as standard deviation of differences.

Any suggestions would be appreciated.

Thanks

On Wed, Jun 11, 2008 at 6:25 AM, Berk Geveci <berk.geveci at kitware.com> wrote:
> Yes, there is a better way of doing this: using the python
> programmable filter. What you will need:
>
> 1. cvs paraview or a development snapshot from here:
> http://paraview.org/New/download.html
>
> 2. numpy
>
> We are not distributing numpy with our binaries yet so your best bet
> is to download python and numpy and then download the ParaView source
> from cvs and compile against that. We will probably start distributing
> numpy with our binaries in the future to avoid this.
>
> Then, you write a scipt like the following:
>
> from paraview import numpy_support
> import numpy
>
> # get the two inputs - note this requires connecting two inputs to the filter
> input0 = self.GetInputDataObject(0, 0)
> input1 = self.GetInputDataObject(0, 1)
>
> # convert the input arrays to numpy arrays. substitute the right cell/point data
> # and the right array names
> ar0 = numpy_support.vtk_to_numpy(input0.GetPointData().GetArray("Elevation"))
> ar1 = numpy_support.vtk_to_numpy(input1.GetPointData().GetArray("Result"))
>
> # take the difference and convert to vtk array
> out = numpy_support.numpy_to_vtk(ar1 - ar0, 1)
> # give it a name
> out.SetName("difference")
>
> # add it to the output
> self.GetOutputDataObject(0).GetPointData().AddArray(out)
>
>
> In the future, this will be much simpler. My plan is that the above
> will look like:
>
> out = inputs[1].pointdata.Elevation - inputs[0].pointdata.Result
> output.add_array(out, "difference")
>
> I hope this helps.
>
> -berk
>
> On Tue, Jun 10, 2008 at 8:52 PM, amayt <tyamamayt at gmail.com> wrote:
>> Hello,
>>
>> When comparing datasets, you can use the following procedures to
>> integrate multiple data.  However, the resulting data contains
>> multiple data arrays in addition to the original data, which increases
>> the data size.  For example, the data 'h0' is a copy of 'h' in the
>> following case, so it seems like that 'h' is no longer necessary.  Is
>> it possible to remove such data arrays?
>>
>> Any other suggestions to decrease the data size would be appreciated.
>>
>>> 1) Use Calculator #0 to rename the variable from dataset #0
>>>   (eg pick just 'h' with result array name 'h0')
>>> 2) Use Calculator #1 to rename the variable from dataset #1
>>>   (eg 'h' to 'h1')
>>> 3) Use "Append Attributes" while having selected both Calculator #0
>>>   and #1 to combine the output from 1) and 2). AppendAttributes1
>>>   now contains 'h0', 'h1' as well as 'h' (which one?).
>>> 4) Use Calculator #2 on AppendAttributes1 to take difference (eg h1-h0).
>> _______________________________________________
>> ParaView mailing list
>> ParaView at paraview.org
>> http://www.paraview.org/mailman/listinfo/paraview
>>
>