Re: [Matplotlib-users] Making the Axes Plot Area TransparentRatherthan having Color

[Charlie M. <cw...@gm...>]

On 8/4/06, Richard House <rm...@rh...> wrote:
> -----Original Message-----
> On Behalf Of Charlie Moad, Sent: 04 August 2006 14:34
> Subject: Re: [Matplotlib-users] Making the Axes Plot Area TransparentRatherthan having Color
>
> On 8/4/06, Richard House <rm...@rh...> wrote:
> > -----Original Message-----
> >  On Behalf Of Charlie Moad, Sent: 04 August 2006 13:59
> > Subject: Re: [Matplotlib-users] Making the Axes Plot Area Transparent
> > Ratherthan having Color
> >
> > On 8/4/06, Richard House <rm...@rh...> wrote:
> > > I am trying to make a transparent chart and have managed to make the frame area transparent using frameon(False). However I can't find out how to make the plot area background on the chart transparent, without using a graphics package to process the matplotlib produced image. I have searched the mailing list and documentation but haven't found any information on the subject.
> > > Can anyone offer any advice?
> >
> > gca().axesPatch.set_alpha(0.0)
> >
> > Thanks for that Charlie. It works, but removes the axes plot area border so the x and y axes have no solid line. Is there a simple way to avoid that or do I need to plot the axes as lines?
>
> I don't think I am seeing that effect.  I am attaching my output for this:
>
> figure(frameon=False)
> plot(rand(100))
> gca().axesPatch.set_alpha(0.0)
> savefig('out.png')
>
> Thanks again Charlie. Just had a few experiments. I wonder if we are dealing with difference platform/version behaviour? I was running WinXP, Python 2.4, Matplotlib 0.82 and your example code did not produce transparency for me. However the following code did but without the plot area border (example attached):
> #!/usr/bin/env python
> from pylab import *
> fig = figure()
> fig.set_frameon(False)
> plot(rand(100))
> gca().axesPatch.set_alpha(0.0)
> savefig('transparent.png')
>
> On upgrading to 0.87.3 (0.87.4 complains about numpy versions), and running this code, the png file produced does include a border, the same as in your output sample. The upgraded matplotlib also produces a plot area border in my own chart code, so the problem is solved.

Sounds like you got it working then.  Mpl 0.87.4 requires numpy-0.9.8.
 I think 0.87.3 requires 0.9.6, but I am starting to forget all the
number combinations.  Once numpy 1.0 is released, this mess should be
gone.