24

How can I flip the origin of a matplotlib plot to be in the upper-left corner - as opposed to the default lower-left? I'm using matplotlib.pylab.plot to produce the plot (though if there is another plotting routine that is more flexible, please let me know).

I'm looking for the equivalent of the matlab command: axis ij;

Also, I've spent a couple hours surfing matplotlib help and google but haven't come up with an answer. Some info on where I could have looked up the answer would be helpful as well.

tshepang
  • 12,111
  • 21
  • 91
  • 136
Nate
  • 768
  • 3
  • 9
  • 17

5 Answers5

32

The easiest way is to use:

plt.gca().invert_yaxis()

After you plotted the image. Origin works only for imshow.

G M
  • 20,759
  • 10
  • 81
  • 84
  • 1
    This worked great when putting maps on satellite imagery with Cartopy. Note that, at least in my use case, this has to run AFTER the `imshow()`.Thanks! – kevinmicke Feb 14 '18 at 23:24
  • In case you're using `subplots` and `sharex`/`sharey`, then call this function only once otherwise for even number of cols/rows, you'll end up in the unchanged way again. – pas-calc Mar 23 '23 at 12:11
13

axis ij just makes the y-axis increase downward instead of upward, right? If so, then matplotlib.axes.invert_yaxis() might be all you need -- but I can't test that right now.

If that doesn't work, I found a mailing post suggesting that

setp(gca(), 'ylim', reversed(getp(gca(), 'ylim')))

might do what you want to resemble axis ij.

Mark Rushakoff
  • 249,864
  • 45
  • 407
  • 398
  • I never could get invert_yaxis() to work. If it did, that would be ideal. From what I could piece together, the setp command should work, but "reversed" returns an iterator. I couldn't get it to return a sequence instead, though I am still relatively new to this. So it may be easy. – Nate Aug 31 '09 at 18:35
  • "`axis ij` just makes the y-axis increase downward instead of upward, right?" That is/was not always true: "`axis ij` places the coordinate system origin in the upper-left corner. The `i`-axis is vertical, with values increasing from top to bottom. The `j`-axis is horizontal with values increasing from left to right." (http://www.ece.northwestern.edu/local-apps/matlabhelp/techdoc/ref/axis.html) – bers Jun 29 '22 at 15:03
10

For an image or contour plot, you can use the keyword origin = None | 'lower' | 'upper' and for a line plot, you can set the ylimits high to low.

from pylab import *
A = arange(25)/25.
A = A.reshape((5,5))

figure()
imshow(A, interpolation='nearest', origin='lower')

figure()
imshow(A, interpolation='nearest')

d = arange(5)
figure()
plot(d)
ylim(5, 0)

show()
tom10
  • 67,082
  • 10
  • 127
  • 137
  • 1
    Directly setting the axis limits with "ylim([y1,y2])" or "axis([x1,x2,y1,y2])" worked will. It's not completely generic, but I'm sure I can figure something out to make it generic. Maybe the setp command suggested above. Thanks for the help! – Nate Aug 31 '09 at 18:38
7

The following is a basic way to achieve this

ax=pylab.gca()

ax.set_ylim(ax.get_ylim()[::-1])
Dungeon Hunter
  • 19,827
  • 13
  • 59
  • 82
Peter Caldwell
  • 71
  • 1
  • 1
6

This

plt.ylim(max(plt.ylim()), min(plt.ylim()))

has an advantage over this

plt.gca().invert_yaxis()

and is that if you are in interactive mode and you repeatedly plot the same plot (maybe with updated data and having a breakpoint after the plot) the y axis won't keep inverting every time.

martinako
  • 2,690
  • 1
  • 25
  • 44