[prev in list] [next in list] [prev in thread] [next in thread]
List: vtk-developers
Subject: Re: [vtk-developers] Problem with using vtkArrayCalculator to extract
From: Randall Hand <randall.hand () gmail ! com>
Date: 2005-09-14 20:55:33
Message-ID: b0226472050914135560cc88de () mail ! gmail ! com
[Download RAW message or body]
Well, I'm building a vtkDoubleArray, and the vtkArrayCalculator claims it's
returning a vtkDoubleArray.. So that shouldn't be the case in this
situation.
On 9/14/05, Berk Geveci <berk.geveci@gmail.com> wrote:
>
> This is a guess but if the array calculator outputs a data type than
> the input's scalars, no imaging filter can be used down the pipeline.
> This is a known bug. Essentially no (non-imaging) filter that changes
> the data type of the scalars is allowed in a pipeline with imaging
> filters. There is work (slow) going on to fix this issue. I would not
> hold my breath for it's completion.
>
> On 9/14/05, Randall Hand <randall.hand@gmail.com> wrote:
> > Ok, I posted this once before, but here it is again with some more
> > information.
> >
> > I've written a piece of code to create a synthetic dataset. It's a 128^3
> > dataset, each point with a 3-component double value consisting of
> (sin(x),
> > junk, morejunk). I extract just the first component, and do an FFT.
> You'll
> > find the code at the end of this message, along with my
> "PrintStatistics"
> > function attached.
> >
> > You'll see from the #if/#endif blocks that there's two possible ways to
> do
> > it. The first is with a vtkArrayCalculator, where I set a very trivial
> > function to simply copy out the one value. The other way is with a
> > vtkImageExtractComponents. When doing it with the vtkArrayCalculator,
> the
> > filter seems to work just fine but then the FFT yields values like :
> >
> > p Array [0]: vtkDoubleArray "result", 2,097,152 points, 8 bytes per
> > point
> > p Magnitude Range: 0.0000 1482912.5793
> > p Component 0 Range: -1048575.9848
> > 1048575.8760
> > p Component 1 Range: -1048572.9243
> > 1048579.0968
> > With a simple sin wave, the result should be almost purely imaginary, so
> > the Component 0 range should be almost 0 to 0.
> >
> > When I switch to using the vtkImageExtractComponents, the results look
> > better:
> >
> >
> > p Array [0]: vtkDoubleArray "image", 2,097,152 points, 8 bytes per
> > point
> > p Magnitude Range: 0.0000 1048576.0091
> > p Component 0 Range: -0.0557 0.0011
> > p Component 1 Range: -1048576.0089
> > 1048576.0091
> >
> >
> > Unfortunately, when working with the vtkImageExtractComponents filter, I
> > lose the other two fields if I had planned on using them for coloring or
> > something. Also, my results aren't quite what I expected. Much better,
> but
> > on more complex datasets I seem to get results that are a little off
> from
> > the expected result. I don't know if the cause of all this is related,
> but
> > I'm chasing down what I can.
> >
> > So, any ideas?
> >
> > ----- Code starts here -------
> >
> > #include <stdio.h>
> > #include <string.h>
> > #include <stdlib.h>
> > #include <unistd.h>
> > #include "stat.h"
> > #include <sys/types.h>
> > #include <sys/stat.h>
> > #include <sys/mman.h>
> > #include <fcntl.h>
> > #include <vtkObject.h>
> > #include "vtkImageData.h"
> > #include "vtkPointData.h"
> > #include <vtkDataArray.h>
> > #include <vtkUnsignedCharArray.h>
> > #include <vtkDoubleArray.h>
> > #include <vtkImageData.h>
> > #include <vtkImageFFT.h>
> > #include <vtkArrayCalculator.h>
> > #include <vtkImageExtractComponents.h>
> >
> > int main (int argc, char *argv[])
> > {
> > double *data;
> > int xdim, ydim, zdim;
> > double kx;
> > int x,y,z;
> > long index;
> > xdim = 128;
> > ydim = 128;
> > zdim = 128;
> >
> > if ((data =
> > (double*)malloc(sizeof(double)*xdim*ydim*zdim*3)) == NULL)
> > {
> > perror("Unable to allocate memory!\n\t");
> > return -1;
> > }
> >
> > printf("Constructing sin(kx * x) dataset...\n");
> > kx = 2.0 * 3.1415926 / (double)xdim;
> > for(x=0; x<xdim; x++) {
> > for(y=0; y<ydim; y++) {
> > for(z=0; z<zdim; z++) {
> > index = z + (y*zdim) + (x * ydim * zdim);
> > data[(index*3)+0] = sin(kx * (double)x);
> > data[(index*3)+1] = cos(kx * (double)x);
> > data[(index*3)+2] = tan(kx * (double)x);
> > }
> > }
> > }
> >
> > vtkDoubleArray *ucPointer = vtkDoubleArray::New();
> > ucPointer->SetNumberOfComponents(3);
> > ucPointer->SetArray(data, xdim*ydim*zdim*3, 1);
> > ucPointer->SetName("image");
> >
> > vtkImageData *image = vtkImageData::New();
> > image->Initialize();
> > image->SetDimensions(xdim, ydim, zdim);
> > image->SetExtent(1, xdim, 1, ydim, 1, zdim);
> > image->SetScalarTypeToDouble();
> > image->SetNumberOfScalarComponents(3);
> > image->GetPointData()->SetScalars(ucPointer);
> > PrintStatistics(image);
> > printf("Extracting Velocity X-Component...\n");
> >
> > #if 0
> > vtkArrayCalculator *extract = vtkArrayCalculator::New();
> > extract->SetInput(image);
> > extract->SetResultArrayName("result");
> > extract->AddScalarVariable("x", "image", 0);
> > extract->SetFunction("(x*1.0)+0.0");
> > extract->ReplaceInvalidValuesOff();
> > extract->Update();
> >
> > extract->GetOutput()->GetPointData()->RemoveArray("image");
> >
> > extract->GetOutput()->GetPointData()->SetActiveScalars("result");
> > #else
> > vtkImageExtractComponents *extract =
> > vtkImageExtractComponents::New();
> > extract->SetInput(image);
> > extract->SetComponents(0);
> > extract->Update();
> > #endif
> > PrintStatistics(extract->GetOutput());
> >
> > printf("Computing FFT...\n");
> > vtkImageFFT *fft = vtkImageFFT::New();
> > fft->SetInput(extract->GetOutput());
> > fft->Update();
> > PrintStatistics(fft->GetOutput());
> >
> > }
> >
> >
> > --
> > Randall Hand
> > Visualization Scientist,
> > ERDC-MSRC Vicksburg, MS
> > Homepage: http://www.yeraze.com
> > _______________________________________________
> > vtk-developers mailing list
> > vtk-developers@vtk.org
> > http://www.vtk.org/mailman/listinfo/vtk-developers
> >
> >
> >
> >
>
--
Randall Hand
Visualization Scientist,
ERDC-MSRC Vicksburg, MS
Homepage: http://www.yeraze.com
[Attachment #3 (text/html)]
Well, I'm building a vtkDoubleArray, and the vtkArrayCalculator claims
it's returning a vtkDoubleArray.. So that shouldn't be the case in this
situation.<br><br><div><span class="gmail_quote">On 9/14/05, <b \
class="gmail_sendername">Berk Geveci</b> <<a \
href="mailto:berk.geveci@gmail.com">berk.geveci@gmail.com</a>> \
wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, \
204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> This is a guess but if the \
array calculator outputs a data type than<br>the input's scalars, no imaging filter \
can be used down the pipeline.<br>This is a known bug. Essentially no (non-imaging) \
filter that changes<br>the data type of the scalars is allowed in a pipeline with \
imaging <br>filters. There is work (slow) going on to fix this issue. I would \
not<br>hold my breath for it's completion.<br><br>On 9/14/05, Randall Hand <<a \
href="mailto:randall.hand@gmail.com">randall.hand@gmail.com</a>> wrote: <br>> \
Ok, I posted this once before, but here it is again with some more<br>> \
information.<br>><br>> I've written a piece of code to create a \
synthetic dataset. It's a 128^3<br>> dataset, each point with a \
3-component double value consisting of (sin(x), <br>> junk, \
morejunk). I extract just the first component, and do an \
FFT. You'll<br>> find the code at the end of this message, along with \
my "PrintStatistics"<br>> function \
attached.<br>><br>> You'll see from the #if/#endif blocks that \
there's two possible ways to do <br>> it. The first is with a \
vtkArrayCalculator, where I set a very trivial<br>> function to simply copy out \
the one value. The other way is with a<br>> \
vtkImageExtractComponents. When doing it with the vtkArrayCalculator, the \
<br>> filter seems to work just fine but then the FFT yields values like \
:<br>><br>> p Array [0]: vtkDoubleArray \
"result", 2,097,152 points, 8 bytes per<br>> \
point<br>> p
Magnitude
Range: 0.0000 1482912.5793<b \
r>> p
Component 0 Range: -1048575.9848<br>> \
1048575.8760<br>> p
Component 1 Range: -1048572.9243<br>> \
1048579.0968<br>> With a simple sin wave, the result should be almost \
purely imaginary, so<br>> the Component 0 range should be almost 0 to \
0.<br>><br>> When I switch to using the vtkImageExtractComponents, \
the results look <br>> better:<br>><br>><br>> \
p Array [0]: vtkDoubleArray "image", \
2,097,152 points, 8 bytes per<br>> \
point<br>> p
Magnitude
Range: 0.0000 1048576.0091<b \
r>> p
Component 0 Range: -0.0557 \
0.0011<br>> p
Component 1 Range: -1048576.0089<br>> \
1048576.0091<br>><br>><br>> Unfortunately, when working with the \
vtkImageExtractComponents filter, I<br>> lose the other two fields if I had \
planned on using them for coloring or <br>> something. Also, my results \
aren't quite what I expected. Much better, but<br>> on more complex \
datasets I seem to get results that are a little off from<br>> the expected \
result. I don't know if the cause of all this is related, but <br>> I'm \
chasing down what I can.<br>><br>> So, any \
ideas?<br>><br>> ----- Code starts here -------<br>><br>> \
#include <stdio.h><br>> #include \
<string.h><br>> #include <stdlib.h \
><br>> #include <unistd.h><br>> #include \
"stat.h"<br>> #include \
<sys/types.h><br>> #include \
<sys/stat.h><br>> #include \
<sys/mman.h><br>> #include <fcntl.h> \
<br>> #include <vtkObject.h><br>> #include \
"vtkImageData.h"<br>> #include \
"vtkPointData.h"<br>> #include \
<vtkDataArray.h><br>> #include <vtkUnsignedCharArray.h><br> \
> #include <vtkDoubleArray.h><br>> #include \
<vtkImageData.h><br>> #include \
<vtkImageFFT.h><br>> #include \
<vtkArrayCalculator.h><br>> #include \
<vtkImageExtractComponents.h> <br>><br>> int main (int argc, \
char *argv[])<br>> {<br>> double \
*data;<br>> int xdim, ydim, \
zdim;<br>> double \
kx;<br>> int \
x,y,z;<br>> long \
index;<br>> xdim = 128; \
<br>> ydim = \
128;<br>> zdim = \
128;<br>><br>> if ((data =<br>> \
(double*)malloc(sizeof(double)*xdim*ydim*zdim*3)) == NULL)<br>> \
{<br>> perror("Unable \
to allocate memory!\n\t"); \
<br>> return \
-1;<br>> }<br>><br>> printf("Constructing \
sin(kx * x) dataset...\n");<br>> kx = 2.0 \
* 3.1415926 / (double)xdim;<br>> for(x=0; \
x<xdim; x++) {<br>> for(y=0; \
y<ydim; y++) { <br>> for(z=0; \
z<zdim; z++) {<br>> index
= z + (y*zdim) + (x * ydim * \
zdim);<br>> data[(index*3)+0]
= sin(kx * (double)x);<br>> data[(index*3)+1]
= cos(kx * (double)x);<br>> data[(index*3)+2]
= tan(kx * (double)x);<br>> & \
nbsp; }<br>> \
}<br>> }<br>><br>> vtkDoubleArray \
*ucPointer = vtkDoubleArray::New();<br>> ucPoint \
er->SetNumberOfComponents(3);<br>> ucPointer->SetArray(data, \
xdim*ydim*zdim*3, 1); \
<br>> ucPointer->SetName("image");<br>><br>> vtkImageData \
*image = vtkImageData::New();<br>> image->Ini \
tialize();<br>> image->SetDimensions(xdim, \
ydim, zdim);<br>> image->SetExtent(1, xdim, \
1, ydim, 1, zdim); <br>> image->SetScalarType \
ToDouble();<br>> image->SetNumberOfScalarComp \
onents(3);<br>> image->GetPointData()->Set \
Scalars(ucPointer);<br>> PrintStatistics(image);<br>> printf("Extracting \
Velocity X-Component...\n"); <br>><br>> #if \
0<br>> vtkArrayCalculator *extract = \
vtkArrayCalculator::New();<br>> extract->SetI \
nput(image);<br>> extract->SetResultArrayName \
("result");<br>> extract->AddScalarVariable("x", \
"image", 0); \
<br>> extract->SetFunction("(x*1.0)+0.0& \
quot;);<br>> extract->ReplaceInvalidValuesOff \
();<br>> extract->Update();<br>><br>> \
extract->GetOutput()->GetPointData()->RemoveArray("image"); \
<br>><br>> extract->GetOutput()->GetPointData()->SetActiveScalars(" \
;result");<br>> #else<br>> vtkImageExtractComponents \
*extract =<br>> vtkImageExtractComponents::New();<br>> extract->SetInput(image);
<br>> extract->SetComponents(0);<br>>&nbs \
p; extract->Update();<br>> #endif<br>> \
; PrintStatistics(extract->GetOutput());<br>><br>> printf("Computing \
FFT...\n");<br>> vtkImageFFT *fft = \
vtkImageFFT::New(); <br>> fft->SetInput(extra \
ct->GetOutput());<br>> fft->Update();<br>& \
gt; PrintStatistics(fft->GetOutput());<br>><br>> }<br>><br>><br>> \
--<br>> Randall Hand<br>> Visualization Scientist, <br>> ERDC-MSRC \
Vicksburg, MS<br>> Homepage: <a \
href="http://www.yeraze.com">http://www.yeraze.com</a><br>> \
_______________________________________________<br>> vtk-developers mailing \
list<br>> <a href="mailto:vtk-developers@vtk.org"> \
vtk-developers@vtk.org</a><br>> <a \
href="http://www.vtk.org/mailman/listinfo/vtk-developers">http://www.vtk.org/mailman/l \
istinfo/vtk-developers</a><br>><br>><br>><br>><br></blockquote></div><br><br \
clear="all"> <br>-- <br>Randall Hand<br>Visualization Scientist, <br>ERDC-MSRC \
Vicksburg, MS<br>Homepage: <a href="http://www.yeraze.com">http://www.yeraze.com</a>
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic