16bit tiff support?

Community matplotlib-users
1

Hi,

is there a backend that supports 16bit tiff images?
According to the website GDK supports tiff but that's wrong:

>>>import matplotlib
>>>matplotlib.use('GDK')
>>>import matplotlib.pyplot as pyplot
>>>pyplot.imsave(arr=X, fname='test.tif')
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib/pymodules/python2.6/matplotlib/pyplot.py", line 1425, in imsave
return _imsave(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/image.py", line 813, in imsave
fig.savefig(fname, dpi=1, format=format)
File "/usr/lib/pymodules/python2.6/matplotlib/figure.py", line 1033, in savefig
self.canvas.print_figure(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/backend_bases.py", line 1420, in print_figure
'%s.' % (format, ', '.join(formats)))
ValueError: Format "tif" is not supported.
Supported formats: emf, eps, pdf, png, ps, raw, rgba, svg, svgz.
>>>matplotlib.backends.backend
'gdk'

matplotlib 0.99.0 python 2.6.4 ubuntu karmic x64

If matplotlib cannot provide tiff support, does someone know an alternative? PIL doesn't work either, at least not intuitively.

Cheers

Philipp

···

--
Philipp Lies

Max Planck Institute for Biological Cybernetics
Computational Vision & Neuroscience Group
Spemannstr. 41
D-72076 Tuebingen
Germany

Phone: +49-7071-601-1788
Fax: +49-7071-601-552
E-Mail: philipp.lies@...368...
http://www.kyb.mpg.de/bethgegroup

2

Philipp Lies wrote:

Hi,

is there a backend that supports 16bit tiff images?

Can you just use png, and use the netpbm utilities or ImageMagick convert program to go to and from tiff?

Eric

···

According to the website GDK supports tiff but that's wrong:

>>>import matplotlib
>>>matplotlib.use('GDK')
>>>import matplotlib.pyplot as pyplot
>>>pyplot.imsave(arr=X, fname='test.tif')
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib/pymodules/python2.6/matplotlib/pyplot.py", line 1425, in imsave
return _imsave(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/image.py", line 813, in imsave
fig.savefig(fname, dpi=1, format=format)
File "/usr/lib/pymodules/python2.6/matplotlib/figure.py", line 1033, in savefig
self.canvas.print_figure(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/backend_bases.py", line 1420, in print_figure
'%s.' % (format, ', '.join(formats)))
ValueError: Format "tif" is not supported.
Supported formats: emf, eps, pdf, png, ps, raw, rgba, svg, svgz.
>>>matplotlib.backends.backend
'gdk'

matplotlib 0.99.0 python 2.6.4 ubuntu karmic x64

If matplotlib cannot provide tiff support, does someone know an alternative? PIL doesn't work either, at least not intuitively.

Cheers

Philipp

3

Philipp Lies wrote:

Hi,

is there a backend that supports 16bit tiff images?

Can you just use png, and use the netpbm utilities or ImageMagick
convert program to go to and from tiff?

Would be 'dirty' but acceptable if matplotlib would support saving uncompressed grayscale uint16 png files. But saving nxm uint16 arrays leads to nxmx3 float arrays which do not even closely resemble my original data.
Example:
A
array([[47705, 11865, 739, 16941, 37700],
        [64321, 26860, 49945, 63556, 13498],
        [ 2676, 7720, 5995, 22399, 32735],
        [56577, 34443, 6636, 23409, 61331],
        [ 1020, 26013, 34677, 37262, 36136]], dtype=uint16)
imsave('t.png',A)
B = imread('t.png')
B[:,:,0]

array([[ 1., 0., 0., 0., 0.74117649],
        [ 0.49803922, 0.19607843, 1., 0.5529412 , 0. ],
        [ 0., 0., 0., 0., 0.48627451],
        [ 1., 0.57647061, 0., 0.01960784, 0.71372551],
        [ 0., 0.14509805, 0.58823532, 0.72941178, 0.66666669]], dtype=float32)

···

On 02/12/2010 07:49 PM, Eric Firing wrote:

According to the website GDK supports tiff but that's wrong:

>>>import matplotlib
>>>matplotlib.use('GDK')
>>>import matplotlib.pyplot as pyplot
>>>pyplot.imsave(arr=X, fname='test.tif')
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib/pymodules/python2.6/matplotlib/pyplot.py", line 1425,
in imsave
return _imsave(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/image.py", line 813, in
imsave
fig.savefig(fname, dpi=1, format=format)
File "/usr/lib/pymodules/python2.6/matplotlib/figure.py", line 1033,
in savefig
self.canvas.print_figure(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/backend_bases.py", line
1420, in print_figure
'%s.' % (format, ', '.join(formats)))
ValueError: Format "tif" is not supported.
Supported formats: emf, eps, pdf, png, ps, raw, rgba, svg, svgz.
>>>matplotlib.backends.backend
'gdk'

matplotlib 0.99.0 python 2.6.4 ubuntu karmic x64

If matplotlib cannot provide tiff support, does someone know an
alternative? PIL doesn't work either, at least not intuitively.

Cheers

Philipp

--
Philipp Lies

Max Planck Institute for Biological Cybernetics
Computational Vision & Neuroscience Group
Spemannstr. 41
D-72076 Tuebingen
Germany

Phone: +49-7071-601-1788
Fax: +49-7071-601-552
E-Mail: philipp.lies@...368...
http://www.kyb.mpg.de/bethgegroup

4

Philipp Lies wrote:

  

Philipp Lies wrote:
    

Hi,

is there a backend that supports 16bit tiff images?
      
The macosx backend supports tiff.

-Jeff

···

On 02/12/2010 07:49 PM, Eric Firing wrote:

Can you just use png, and use the netpbm utilities or ImageMagick
convert program to go to and from tiff?
    

Would be 'dirty' but acceptable if matplotlib would support saving uncompressed grayscale uint16 png files. But saving nxm uint16 arrays leads to nxmx3 float arrays which do not even closely resemble my original data.
Example:
A
array([[47705, 11865, 739, 16941, 37700],
        [64321, 26860, 49945, 63556, 13498],
        [ 2676, 7720, 5995, 22399, 32735],
        [56577, 34443, 6636, 23409, 61331],
        [ 1020, 26013, 34677, 37262, 36136]], dtype=uint16)
imsave('t.png',A)
B = imread('t.png')
B[:,:,0]

array([[ 1., 0., 0., 0., 0.74117649],
        [ 0.49803922, 0.19607843, 1., 0.5529412 , 0. ],
        [ 0., 0., 0., 0., 0.48627451],
        [ 1., 0.57647061, 0., 0.01960784, 0.71372551],
        [ 0., 0.14509805, 0.58823532, 0.72941178, 0.66666669]], dtype=float32)

According to the website GDK supports tiff but that's wrong:

import matplotlib
matplotlib.use('GDK')
import matplotlib.pyplot as pyplot
pyplot.imsave(arr=X, fname='test.tif')
            

Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib/pymodules/python2.6/matplotlib/pyplot.py", line 1425,
in imsave
return _imsave(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/image.py", line 813, in
imsave
fig.savefig(fname, dpi=1, format=format)
File "/usr/lib/pymodules/python2.6/matplotlib/figure.py", line 1033,
in savefig
self.canvas.print_figure(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/backend_bases.py", line
1420, in print_figure
'%s.' % (format, ', '.join(formats)))
ValueError: Format "tif" is not supported.
Supported formats: emf, eps, pdf, png, ps, raw, rgba, svg, svgz.
      

matplotlib.backends.backend
            

'gdk'

matplotlib 0.99.0 python 2.6.4 ubuntu karmic x64

If matplotlib cannot provide tiff support, does someone know an
alternative? PIL doesn't work either, at least not intuitively.

Cheers

Philipp

5

Jeff Whitaker wrote:

Philipp Lies wrote:

Philipp Lies wrote:
   

Hi,

is there a backend that supports 16bit tiff images?
      
The macosx backend supports tiff.

Thanks, but I need a linux backend :-/

···

On 02/12/2010 07:49 PM, Eric Firing wrote:

Can you just use png, and use the netpbm utilities or ImageMagick
convert program to go to and from tiff?
    

Would be 'dirty' but acceptable if matplotlib would support saving
uncompressed grayscale uint16 png files. But saving nxm uint16 arrays
leads to nxmx3 float arrays which do not even closely resemble my
original data.
Example:
A
array([[47705, 11865, 739, 16941, 37700],
        [64321, 26860, 49945, 63556, 13498],
        [ 2676, 7720, 5995, 22399, 32735],
        [56577, 34443, 6636, 23409, 61331],
        [ 1020, 26013, 34677, 37262, 36136]], dtype=uint16)
imsave('t.png',A)
B = imread('t.png')
B[:,:,0]

array([[ 1., 0., 0., 0., 0.74117649],
        [ 0.49803922, 0.19607843, 1., 0.5529412 , 0. ],
        [ 0., 0., 0., 0., 0.48627451],
        [ 1., 0.57647061, 0., 0.01960784, 0.71372551],
        [ 0., 0.14509805, 0.58823532, 0.72941178, 0.66666669]],
dtype=float32)

According to the website GDK supports tiff but that's wrong:

import matplotlib
matplotlib.use('GDK')
import matplotlib.pyplot as pyplot
pyplot.imsave(arr=X, fname='test.tif')
            

Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib/pymodules/python2.6/matplotlib/pyplot.py", line 1425,
in imsave
return _imsave(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/image.py", line 813, in
imsave
fig.savefig(fname, dpi=1, format=format)
File "/usr/lib/pymodules/python2.6/matplotlib/figure.py", line 1033,
in savefig
self.canvas.print_figure(*args, **kwargs)
File "/usr/lib/pymodules/python2.6/matplotlib/backend_bases.py", line
1420, in print_figure
'%s.' % (format, ', '.join(formats)))
ValueError: Format "tif" is not supported.
Supported formats: emf, eps, pdf, png, ps, raw, rgba, svg, svgz.
     

matplotlib.backends.backend
            

'gdk'

matplotlib 0.99.0 python 2.6.4 ubuntu karmic x64

If matplotlib cannot provide tiff support, does someone know an
alternative? PIL doesn't work either, at least not intuitively.