Re: [Matplotlib-users] large postscript files

[<hu...@ya...>]

Le mercredi 6 septembre 2006 09:49, Darren Dale a =E9crit=A0:
> On Wednesday 06 September 2006 07:49, jo...@st... wrote:
> > Hi,
> >
> > The following plot
> >
> > >>> from numarray import *
> > >>> x =3D arange(80000)
> > >>> from pylab import plot,show
> > >>> plot(x,x)
> > >>> show()
> >
> > and saving in postscript format generated a file of 1.5MB, while the
> > equivalent is only 288KB in xmgrace (another plotting program). If I use
> > plot(x,x,"k,"), this even leads to a horrible 8.0MB. How come? I
> > understand there is an issue with the fonts, but this can't be the only
> > responsible, can it? Plotting no points, just a title, and saving in
> > postscript leads to a file size of only 133KB.
> >
> > FWIW, I'm trying to make postscript plots using Python 2.4.1, latest
> > numarray, and matplotlib 0.83.2.
>
> Each data point consists of a line like
>
> 54.3869 28.6788 l
>
> which is 17 bytes long. 17*80000 =3D 1.36MB. Maybe we dont need as many s=
ig
> figs, that could cut the size down by maybe 25%.
>
> Darren

Another things that can be good to add is a better clipping. For the moment=
=20
there are no clipping that means if I have something like::


x =3D  arange(1000)
plot(x,x)
xlim(30,40)
savefig('test.svg')


All the point are stored in the file, not only the part between 30 and 40.

This is something I spoke before and I fill a bug or feature request on=20
sourceforge (I'have been unable to find it today but I looked fast).

You can check the problem with the svg created if you open it with inkscape=
,=20
you'll see all the point. I understood that this is a bug inside inkscape t=
oo=20
butthat means that all the point are there and it's a big concern for me wh=
en=20
I have to plot plenty of spectra and use only a small range because I know=
=20
that my file at the end will be huge. The only way I found is 1) save in pn=
g=20
and use another software to save it in eps but I loose the fact that eps is=
 a=20
vectoriel format 2) do the clipping myself or 3) create my plot with anothe=
r=20
software like pgplot or gnuplot.=20