Thread: Re: [matplotlib-devel] Misalignment imshow vs. grid lines

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Nicolas

I get that too, (with your script and various things in my work).
But if you zoom in, the lines are in the right place.   Is it
some kind of screen aliasing?

Jules

Thanks for testing.

If I zoom at any line cross, the lines are definitely at the wrong place for me.
As for screen aliasing I'm not sure since both the png and pdf seems to be wrong in my case.
Weird !

Nicolas

On Oct 29, 2012, at 15:40 , jules hummon wrote:

> Nicolas
> 
> I get that too, (with your script and various things in my work).
> But if you zoom in, the lines are in the right place.   Is it
> some kind of screen aliasing?
> 
> Jules
> 
> ------------------------------------------------------------------------------
> The Windows 8 Center - In partnership with Sourceforge
> Your idea - your app - 30 days.
> Get started!
> http://windows8center.sourceforge.net/
> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

On Mon, Oct 29, 2012 at 11:25 AM, Nicolas Rougier
<Nic...@in...>wrote:

>
>
> Thanks for testing.
>
> If I zoom at any line cross, the lines are definitely at the wrong place
> for me.
> As for screen aliasing I'm not sure since both the png and pdf seems to be
> wrong in my case.
> Weird !
>
>
>
Which version of matplotlib are you using, just for reference.

Ben Root

matplotlib 1.2.x
python     2.7.2
osx        10.7.5

Nicolas

On Oct 29, 2012, at 16:29 , Benjamin Root wrote:

> 
> 
> On Mon, Oct 29, 2012 at 11:25 AM, Nicolas Rougier <Nic...@in...> wrote:
> 
> 
> Thanks for testing.
> 
> If I zoom at any line cross, the lines are definitely at the wrong place for me.
> As for screen aliasing I'm not sure since both the png and pdf seems to be wrong in my case.
> Weird !
> 
> 
> 
> Which version of matplotlib are you using, just for reference.
> 
> Ben Root
> 

I also see this in mpl 1.1.0, python 2.7.2 with the iPython Qt console
and also with the WXAgg backend

Gary R.

On 30 October 2012 03:16, Nicolas Rougier <Nic...@in...> wrote:
>
>
> matplotlib 1.2.x
> python     2.7.2
> osx        10.7.5
>
>
> Nicolas
>
> On Oct 29, 2012, at 16:29 , Benjamin Root wrote:
>
>>
>>
>> On Mon, Oct 29, 2012 at 11:25 AM, Nicolas Rougier <Nic...@in...> wrote:
>>
>>
>> Thanks for testing.
>>
>> If I zoom at any line cross, the lines are definitely at the wrong place for me.
>> As for screen aliasing I'm not sure since both the png and pdf seems to be wrong in my case.
>> Weird !
>>
>>
>>
>> Which version of matplotlib are you using, just for reference.
>>
>> Ben Root
>>
>
>
> ------------------------------------------------------------------------------
> The Windows 8 Center - In partnership with Sourceforge
> Your idea - your app - 30 days.
> Get started!
> http://windows8center.sourceforge.net/
> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

On Tue, Oct 30, 2012 at 12:25 AM, Nicolas Rougier
<Nic...@in...> wrote:
>
>
> Thanks for testing.
>
> If I zoom at any line cross, the lines are definitely at the wrong place for me.

As  jules hummon commented, I see lines in right places when I zoom in.

> As for screen aliasing I'm not sure since both the png and pdf seems to be wrong in my case.

It still can be some aliasing-related issue. Note that with
interpolation="nearest", the images are rasterized with given dpi even
if you save the figure as pdf.
The agg backend tries to adjust the location of lines and images so
that they are well-aligned with the pixels, and the issue seems to be
related with that behavior.

In my case, using interpolation="none" worked out okay. So give it a try.

Regards,

-JJ

> Weird !
>
>
> Nicolas
>
>
> On Oct 29, 2012, at 15:40 , jules hummon wrote:
>
>> Nicolas
>>
>> I get that too, (with your script and various things in my work).
>> But if you zoom in, the lines are in the right place.   Is it
>> some kind of screen aliasing?
>>
>> Jules
>>
>> ------------------------------------------------------------------------------
>> The Windows 8 Center - In partnership with Sourceforge
>> Your idea - your app - 30 days.
>> Get started!
>> http://windows8center.sourceforge.net/
>> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
>> _______________________________________________
>> Matplotlib-devel mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>
>
> ------------------------------------------------------------------------------
> The Windows 8 Center - In partnership with Sourceforge
> Your idea - your app - 30 days.
> Get started!
> http://windows8center.sourceforge.net/
> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

You're right. Using 'none' interpolation seems to solve the problem. Good to know !

One last question, why is the 'none' interpolation restricted to Agg/PS/pdf ?

Nicolas

On Oct 30, 2012, at 6:53 , Jae-Joon Lee wrote:

> On Tue, Oct 30, 2012 at 12:25 AM, Nicolas Rougier
> <Nic...@in...> wrote:
>> 
>> 
>> Thanks for testing.
>> 
>> If I zoom at any line cross, the lines are definitely at the wrong place for me.
> 
> As  jules hummon commented, I see lines in right places when I zoom in.
> 
>> As for screen aliasing I'm not sure since both the png and pdf seems to be wrong in my case.
> 
> It still can be some aliasing-related issue. Note that with
> interpolation="nearest", the images are rasterized with given dpi even
> if you save the figure as pdf.
> The agg backend tries to adjust the location of lines and images so
> that they are well-aligned with the pixels, and the issue seems to be
> related with that behavior.
> 
> In my case, using interpolation="none" worked out okay. So give it a try.
> 
> Regards,
> 
> -JJ
> 
> 
>> Weird !
>> 
>> 
>> Nicolas
>> 
>> 
>> On Oct 29, 2012, at 15:40 , jules hummon wrote:
>> 
>>> Nicolas
>>> 
>>> I get that too, (with your script and various things in my work).
>>> But if you zoom in, the lines are in the right place.   Is it
>>> some kind of screen aliasing?
>>> 
>>> Jules
>>> 
>>> ------------------------------------------------------------------------------
>>> The Windows 8 Center - In partnership with Sourceforge
>>> Your idea - your app - 30 days.
>>> Get started!
>>> http://windows8center.sourceforge.net/
>>> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
>>> _______________________________________________
>>> Matplotlib-devel mailing list
>>> Mat...@li...
>>> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>> 
>> 
>> ------------------------------------------------------------------------------
>> The Windows 8 Center - In partnership with Sourceforge
>> Your idea - your app - 30 days.
>> Get started!
>> http://windows8center.sourceforge.net/
>> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
>> _______________________________________________
>> Matplotlib-devel mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

Sorry, I was too fast in my reply.

When I save the figure, the png output is ok while the pdf is displaying some kind of interpolation with the image.

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

n = 16
fig = plt.figure(figsize=(6,6))
Z = np.array(([0,1]*(n//2) + [1,0]*(n//2))*(n//2)).reshape(n,n)
plt.imshow(Z, interpolation='none', cmap=plt.cm.gray, extent=[0,n,0,n], alpha=.25)
plt.xticks(np.arange(0,n), [])
plt.yticks(np.arange(0,n), [])
plt.grid(ls='solid')
plt.savefig('pylab-grid.png')
plt.savefig('pylab-grid.pdf')
plt.show()

Just for the record, "Skim", "Preview" and "Adobe Reader" on OSX do not give the same output on the saved pdf.
"Adobe Reader" displays the expected result (same as saved png) while "Preview" and "Skim" are apparently trying to make some (bad) interpolation of the checkboard image.

Nicolas

On Oct 30, 2012, at 7:52 , Nicolas Rougier wrote:

> 
> 
> You're right. Using 'none' interpolation seems to solve the problem. Good to know !
> 
> One last question, why is the 'none' interpolation restricted to Agg/PS/pdf ?
> 
> 
> Nicolas
> 
> 
> 
> On Oct 30, 2012, at 6:53 , Jae-Joon Lee wrote:
> 
>> On Tue, Oct 30, 2012 at 12:25 AM, Nicolas Rougier
>> <Nic...@in...> wrote:
>>> 
>>> 
>>> Thanks for testing.
>>> 
>>> If I zoom at any line cross, the lines are definitely at the wrong place for me.
>> 
>> As  jules hummon commented, I see lines in right places when I zoom in.
>> 
>>> As for screen aliasing I'm not sure since both the png and pdf seems to be wrong in my case.
>> 
>> It still can be some aliasing-related issue. Note that with
>> interpolation="nearest", the images are rasterized with given dpi even
>> if you save the figure as pdf.
>> The agg backend tries to adjust the location of lines and images so
>> that they are well-aligned with the pixels, and the issue seems to be
>> related with that behavior.
>> 
>> In my case, using interpolation="none" worked out okay. So give it a try.
>> 
>> Regards,
>> 
>> -JJ
>> 
>> 
>>> Weird !
>>> 
>>> 
>>> Nicolas
>>> 
>>> 
>>> On Oct 29, 2012, at 15:40 , jules hummon wrote:
>>> 
>>>> Nicolas
>>>> 
>>>> I get that too, (with your script and various things in my work).
>>>> But if you zoom in, the lines are in the right place.   Is it
>>>> some kind of screen aliasing?
>>>> 
>>>> Jules
>>>> 
>>>> ------------------------------------------------------------------------------
>>>> The Windows 8 Center - In partnership with Sourceforge
>>>> Your idea - your app - 30 days.
>>>> Get started!
>>>> http://windows8center.sourceforge.net/
>>>> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
>>>> _______________________________________________
>>>> Matplotlib-devel mailing list
>>>> Mat...@li...
>>>> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>>> 
>>> 
>>> ------------------------------------------------------------------------------
>>> The Windows 8 Center - In partnership with Sourceforge
>>> Your idea - your app - 30 days.
>>> Get started!
>>> http://windows8center.sourceforge.net/
>>> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
>>> _______________________________________________
>>> Matplotlib-devel mailing list
>>> Mat...@li...
>>> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
> 
> 
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

Thanks for making this so easy to reproduce. It is so much easier when
there is a small, self contained, correct example!

Initially I was thinking that the problem was some kind of snapping issue
with the ticks, but it turns out, if you change the interpolation scheme to
"none", you get perfect alignment:

import numpy as np
import matplotlib.pyplot as plt

n = 16
fig = plt.figure(figsize=(10,10))

Z = np.zeros((n, n))
Z[::2, ::2] = 1
Z[1::2, 1::2] = 1

# pick the interpolation you want:
interp = 'nearest'
interp = 'none'

plt.imshow(Z, interpolation=interp,
                cmap='gray', extent=[0, n, 0, n],
                alpha=0.25)

plt.xticks(np.arange(n))
plt.yticks(np.arange(n))
plt.grid(ls='solid')

plt.show()

I haven't considered where to go from this point, but I wanted to let you
know about this option.

Thanks,

Phil

On 30 October 2012 07:10, Nicolas Rougier <Nic...@in...> wrote:

>
>
> Sorry, I was too fast in my reply.
>
> When I save the figure, the png output is ok while the pdf is displaying
> some kind of interpolation with the image.
>
> import numpy as np
> import matplotlib
> matplotlib.use('Agg')
> import matplotlib.pyplot as plt
>
> n = 16
> fig = plt.figure(figsize=(6,6))
> Z = np.array(([0,1]*(n//2) + [1,0]*(n//2))*(n//2)).reshape(n,n)
> plt.imshow(Z, interpolation='none', cmap=plt.cm.gray, extent=[0,n,0,n],
> alpha=.25)
> plt.xticks(np.arange(0,n), [])
> plt.yticks(np.arange(0,n), [])
> plt.grid(ls='solid')
> plt.savefig('pylab-grid.png')
> plt.savefig('pylab-grid.pdf')
> plt.show()
>
>
>
>
>
> Just for the record, "Skim", "Preview" and "Adobe Reader" on OSX do not
> give the same output on the saved pdf.
> "Adobe Reader" displays the expected result (same as saved png) while
> "Preview" and "Skim" are apparently trying to make some (bad) interpolation
> of the checkboard image.
>
>
> Nicolas
>
>
> On Oct 30, 2012, at 7:52 , Nicolas Rougier wrote:
>
> >
> >
> > You're right. Using 'none' interpolation seems to solve the problem.
> Good to know !
> >
> > One last question, why is the 'none' interpolation restricted to
> Agg/PS/pdf ?
> >
> >
> > Nicolas
> >
> >
> >
> > On Oct 30, 2012, at 6:53 , Jae-Joon Lee wrote:
> >
> >> On Tue, Oct 30, 2012 at 12:25 AM, Nicolas Rougier
> >> <Nic...@in...> wrote:
> >>>
> >>>
> >>> Thanks for testing.
> >>>
> >>> If I zoom at any line cross, the lines are definitely at the wrong
> place for me.
> >>
> >> As  jules hummon commented, I see lines in right places when I zoom in.
> >>
> >>> As for screen aliasing I'm not sure since both the png and pdf seems
> to be wrong in my case.
> >>
> >> It still can be some aliasing-related issue. Note that with
> >> interpolation="nearest", the images are rasterized with given dpi even
> >> if you save the figure as pdf.
> >> The agg backend tries to adjust the location of lines and images so
> >> that they are well-aligned with the pixels, and the issue seems to be
> >> related with that behavior.
> >>
> >> In my case, using interpolation="none" worked out okay. So give it a
> try.
> >>
> >> Regards,
> >>
> >> -JJ
> >>
> >>
> >>> Weird !
> >>>
> >>>
> >>> Nicolas
> >>>
> >>>
> >>> On Oct 29, 2012, at 15:40 , jules hummon wrote:
> >>>
> >>>> Nicolas
> >>>>
> >>>> I get that too, (with your script and various things in my work).
> >>>> But if you zoom in, the lines are in the right place.   Is it
> >>>> some kind of screen aliasing?
> >>>>
> >>>> Jules
> >>>>
> >>>>
> ------------------------------------------------------------------------------
> >>>> The Windows 8 Center - In partnership with Sourceforge
> >>>> Your idea - your app - 30 days.
> >>>> Get started!
> >>>> http://windows8center.sourceforge.net/
> >>>>
> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
> >>>> _______________________________________________
> >>>> Matplotlib-devel mailing list
> >>>> Mat...@li...
> >>>> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
> >>>
> >>>
> >>>
> ------------------------------------------------------------------------------
> >>> The Windows 8 Center - In partnership with Sourceforge
> >>> Your idea - your app - 30 days.
> >>> Get started!
> >>> http://windows8center.sourceforge.net/
> >>>
> what-html-developers-need-to-know-about-coding-windows-8-metro-style-apps/
> >>> _______________________________________________
> >>> Matplotlib-devel mailing list
> >>> Mat...@li...
> >>> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
> >
> >
> >
> ------------------------------------------------------------------------------
> > Everyone hates slow websites. So do we.
> > Make your web apps faster with AppDynamics
> > Download AppDynamics Lite for free today:
> > http://p.sf.net/sfu/appdyn_sfd2d_oct
> > _______________________________________________
> > Matplotlib-devel mailing list
> > Mat...@li...
> > https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>