[Paraview] use the same scalar data to create contour and color the surface

Mon Apr 9 09:12:08 EDT 2012

On Sun, Apr 8, 2012 at 9:33 PM, Tom Schoenemann <toms at indiana.edu> wrote:
> Thanks Cory,
>
> I did not know about Load State!  My bad…  I was indeed able to load it.
>
> I still have problems with the data range being truncated, even if I do the calculator step on both files.  For example, on your Wavelet1 and Wavelet2 files, the data ranges are: 37.3531 to 276.829.
>
> This stays the same for the calculator step (highlighting Calculator1 and looking at the Information Tab), as well as the AppendAttributes step (again, as listed in the Information tab).
>
> However, when I do contour, and create the contour (pressing apply), then go to the Information tab, the data range is now 157.091 to 157.091.

This is because I used the default settings for both Wavelet sources,
so they were the same data. Hence, the data range on the contour of
*both* scalars would be the same as the isovalue. If you were to
change one of the settings on the Wavelet source to which you are
*not* applying the contour filter, you should see a larger range of
data values in the information tab for the contour.

> Thus, the range is lost, and the color coding is not what I'd want.  Is this what you also get on your machine?
>
> I've discovered that it has something to do with the Isosurfaces value ranges.  If I leave the default value there (which for your data is 157.091), then the range in the other scalar is truncated to exactly this value: 157.091 (i.e., no range).  But if I "delete all" and then "New range" (and accept the 10 default values it lists), then the range of the remaining scalars extends at least to the next-to-last values on each end of the range.
>
> This makes absolutely no sense to me, but there may be a reason for it?  I don't have enough knowledge of Paraview to know if this is a bug or if it is supposed to do this for some reason.
>
> In any case, for my data, I got it to work by using a binary image of my surface (a brain endocast) and then AppendAttributes this to the stat data file.  For the binary image, I get the full range of stat values if I set the Isosurfaces range value to 1.

My guess is that your stats are computed only inside the binary
segmentation. Contour algorithms such as marching cubes typically
determine where to place vertices on the generated contour by assuming
linear elements in a grid. The result is that when your isovalue is
set to 1, then you get the surface of your segmentation, and the
scalar values interpolated on the contour from your stats scalar field
basically aren't interpolated.When your isovalue is set to 0, the
contour is just about one voxel larger than the boundary between the
inside and the outside of your segmentation. If you don't have any
stats values at the background of your segmentation (or they are
zero), then you shouldn't see anything meaningful in the stats scalar
field computed for the contour. Does that make sense?

In any case, I'm glad you got it to work.

Cory

> It works, but I don't know why.
>
> In any case, thank you very much for helping me with this!!
>
> I'd attach a copy of the file if you want to take a look, but it is 52 MB. I can post it if you are interested, but I seem to get the same odd behavior with your data too.
>
> -Tom
>
>
> On Apr 8, 2012, at 8:39 PM, Cory Quammen wrote:
>
>> On Sun, Apr 8, 2012 at 4:35 PM, Tom Schoenemann <toms at indiana.edu> wrote:
>>> Thanks Cory,
>>>
>>> The file you sent won't open in my Paraview. It does not recognize the
>>> extension (?).
>>
>> That's odd. Did you load it through the File -> Load State menu item?
>> I didn't mention that's what you needed to do to load that file, but
>> in any case, it sounds like you were able to apply my suggestion,
>> albeit not with the outcome I had hoped.
>>
>>> I followed what you suggested, however:
>>>
>>> 1) Loaded the first file and selected "apply".  This file has only binary
>>> scalars (0=background, 1=surface)
>>> 2) Selected calculator and created a function that consisted of:
>>> "scalars*1", to be renamed "surface location"; pressed "apply" again
>>> 3) Loaded the second file and selected "apply"  This file has float scalar
>>> values ranging from 0 to 10.7
>>> 4) control-selected both the calculator and second file, and then selected
>>> "Append Attributes"; and pressed "apply" again.  This resulted in 3 new
>>> items in the pipeline, all called AppendAttributes1.  The one at the bottom
>>> had a box icon (the other two had only arrows).
>>>
>>> When I click on the bottom AppendAttributes1 file and look at the
>>> information tab, it says it has 2 data arrays.  One is a double called
>>> "surface location" (matching what I thought I created in step 2 above), and
>>> has value range of 0 to 1 (which is correct).  However, the other one, which
>>> I assume is the one from the file loaded in step 3 above, now has a range of
>>> only 0 to 1 also (not 0 to 10.7, like it should).  So either it is doing
>>> something odd (like rescaling data??) or it got that data elsewhere?  Why is
>>> it doing this?
>>
>> That is strange. As far as I know, there should be no rescaling or
>> even any recasting performed on the inputs to the Append Attributes
>> filter. At least I can't think of a reason there would need to be;
>> ParaView (actually, the VTK library underneath ParaView) supports
>> images with scalar fields with different component types. One thing to
>> try would be to put a calculator filter on your second image and
>> creating an Append Attributes filter on the outputs of both
>> calculators. The output from the calculator filters consists of double
>> arrays, so if the Append Attributes filter has a problem with
>> different array types, then putting both your images through a
>> calculator filter should fix that. If you are able to post example
>> data, I could give it a try and see what is going on.
>>
>>> If I then compute a contour based on the "surface location" array in
>>> AppendAttributes1, I get the surface I expect to see.
>>>
>>> But the other array data is not any good (it isn't 0 to 10.7 anymore).  Even
>>> if it was, what exactly would I do then to color the surface with this other
>>> array?  Is that done with another filter?  Or is it done in the contour
>>> image?
>>
>> Once you've created the contour, you can select it in the pipeline
>> browser (if it is not already selected), click on the Display tab,
>> find the Color panel, and choose the scalar field by which you want to
>> color the surface from the "Color by" menu.
>>
>> Cory
>>
>>> Thanks for any further help!
>>>
>>> -Tom
>>>
>>> On Apr 8, 2012, at 10:22 AM, Cory Quammen wrote:
>>>
>>> Tom,
>>>
>>> You can load your two images and, assuming they are the same size, use
>>> the Append Attributes filter to combine them. However, chances are
>>> good that the scalars in your image data will have the same name,
>>> meaning that Append attributes will grab only the scalar values from
>>> one of the images. To get around this limitation, you can "rename" the
>>> scalar field in one image by using a Calculator filter on it and
>>> setting the "Result Array Name" to something else. Your pipeline will
>>> look like this *before* applying the Append Attributes filter:
>>>
>>> Image1
>>> ---|
>>>   Calculator1
>>> Image2
>>>
>>> To apply the Append Attributes filter, select the Calculator1 filter
>>> and Image2. Use Ctrl-click on Windows to select more than one object
>>> in the pipeline browser. Then select Append Attributes from the
>>> Filters menu under the Alphabetical submenu.
>>>
>>> Now you should have two scalar fields in an image data. You can use
>>> one to do the contouring and the other to do the pseudo-coloring.
>>>
>>> I've attached a state file with two Wavelet sources in place of image
>>> inputs. It illustrates what I've described here.
>>>
>>> Hope that helps,
>>> Cory
>>>
>>>
>>> On Sun, Apr 8, 2012 at 9:45 AM, Tom Schoenemann <toms at indiana.edu> wrote:
>>>
>>> OK, but is there some way to combine two image files, each with its own
>>>
>>> scalars, into one image?  What I'm thinking about is reading in 2 copies of
>>>
>>> the same image, and using the scalars from one to display the contour, and
>>>
>>> the scalars from the other to display the colors?  I tried doing this, but I
>>>
>>> can't get it to work. I loaded one as the contour, and colored it one color
>>>
>>> only (white).  I then added the copy, an tried to view it as a volume.  It
>>>
>>> let me change the color map parameters, but it seems to display the color
>>>
>>> UNDERNEATH the first one (that has the contour.  I'm sure there must be a
>>>
>>> way to do this, but how?
>>>
>>>
>>> Perhaps there is a way to add 2 scalar fields to one image? Maybe image
>>>
>>> calculator could do this??
>>>
>>>
>>> -Tom
>>>
>>>
>>> On Apr 8, 2012, at 8:30 AM, Cory Quammen wrote:
>>>
>>>
>>> Tom,
>>>
>>>
>>> An isosurface by definition has the same scalar value everywhere on
>>>
>>> the surface, so what you describe happening is exactly what should
>>>
>>> happen, unless I misunderstand. You could contour by one scalar field
>>>
>>> and color by another scalar field and expect to see different colors,
>>>
>>> but if you try to color by the same scalar, you should expect to see
>>>
>>> only one color on the surface.
>>>
>>>
>>> Cory
>>>
>>>
>>> On Sun, Apr 8, 2012 at 3:38 AM, Tom Schoenemann <toms at indiana.edu> wrote:
>>>
>>>
>>> I have a single volume dataset that has scalar values only on a surface, and
>>>
>>>
>>> zeros everywhere else.  The scalar values represent statistics we have
>>>
>>>
>>> calculated for each point on a surface.  I've been able using the contour
>>>
>>>
>>> filter to create an isosurface from this dataset, but I can't get the
>>>
>>>
>>> surface to display anything other than a single color.  I have tried setting
>>>
>>>
>>> "contour by" to "compute normals" and "compute scalars", and then going to
>>>
>>>
>>> "display" and choosing "color by" scalars.  However, only a single value is
>>>
>>>
>>> allowed.  I have tried unselecting "automatically rescale to fit data range"
>>>
>>>
>>> and then entering my own range, but only my minimum value is accepted (even
>>>
>>>
>>> if I enter a different value for the maximum).
>>>
>>>
>>>
>>> Perhaps I am using paraview the wrong way, but I have to say, it isn't
>>>
>>>
>>> intuitive.  Is there a way to use the same scalar data (same file) to both:
>>>
>>>
>>> 1) create a surface, and 2) color code the surface based on the same scalar
>>>
>>>
>>> values?
>>>
>>>
>>>
>>> Thanks for any hints or suggestions.
>>>
>>>
>>>
>>> -Tom
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>>
>>>
>>> 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
>>>
>>>
>>>
>>>
>>>
>>>
>>> --
>>>
>>> Cory Quammen
>>>
>>> Research Associate
>>>
>>> Department of Computer Science
>>>
>>> The University of North Carolina at Chapel Hill
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> --
>>> Cory Quammen
>>> Research Associate
>>> Department of Computer Science
>>> The University of North Carolina at Chapel Hill
>>> <AppendAttributesExample.pvsm>
>>>
>>>
>>> _________________________________________________
>>> P. Thomas Schoenemann
>>>
>>> Associate Professor
>>> Department of Anthropology
>>> Indiana University
>>> Bloomington, IN  47405
>>> Phone: 812-855-8800
>>> E-mail: toms at indiana.edu
>>>
>>> Open Research Scan Archive (ORSA) Co-Director
>>> Consulting Scholar
>>> Museum of Archaeology and Anthropology
>>> University of Pennsylvania
>>>
>>> Homepage: http://mypage.iu.edu/~toms/
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>> --
>> Cory Quammen
>> Research Associate
>> Department of Computer Science
>> The University of North Carolina at Chapel Hill
>
>
>
>

-- 
Cory Quammen
Research Associate
Department of Computer Science
The University of North Carolina at Chapel Hill