Matplotlib savefig does not save axes

Question

I'm trying to save a figure that works fine in IPython inline but does not save the figure to disk with the axes and titles included.

I am using TKAgg backend by default in matplotlibrc.

Any ideas what might be going wrong here? I have clearly set the xlabel and the tick marks work correctly in IPython inline plot.

import matplotlib.pylab as plt  
x = [1,2,3,3]
y = map(lambda(x): x * 2, x)
fig = plt.figure()
ax = fig.add_axes([0,0,1,1])
ax.set_title("bleh")
ax.set_xlabel("xlabel")
ax.plot(x, y, 'r--')
fig.savefig("fig.png")

Lookin

Answer 1

I was having the same problem using Jupyter notebook and the command: %matplotlib notebook. The figure showed correctly in the notebook but didn't print axis and titles when saved with fig.savefig(). I changed %matplotlib notebook to %matplotlib inline and that solved the problem.

Grit

Answer 2

Defining fig = plt.figure(figsize=(15,10)) at the beginning, saving the file as .jpg, and setting bbox_inches='tight' solved the issue for me.

plt.savefig('filename.jpg',bbox_inches='tight', dpi=150)

bbox_inches='tight' seems to fix cropping issues but it didn't work for .png.

Chronopher

In a practical sense this answer was best. I changed only from png to jpg, and the resulting image included the axes that I expected. Good enough for today. But it remains frustrating to me to not understand why these are different and to not understand how to output a .png as intended. – Weave

Molly's answer below explains why ;) – Carlenecarleton

works with .tiff but adding bbox_inches arg – Nmr

Answer 3

Could be facecolor. I work in jupyter lab, and the facecolor default is set to black, so you don't see the axes, even though they are being drawn.

fig = plt.figure(facecolor=(1, 1, 1))

sets the background color to white.

Narceine

this is the correct solution for those having issues in jupyter notebooks. thank you so much! – Coatbridge

Answer 4

You are setting the axis to start at the very bottom left of the figure and to fill up the entire thing. There's no room for the axis labels or the title. Try this:

import matplotlib.pylab as plt  
x = [1,2,3,3]
y = map(lambda(x): x * 2, x)
fig = plt.figure()
ax = fig.add_axes([0.1,0.1,0.75,0.75]) # axis starts at 0.1, 0.1
ax.set_title("bleh")
ax.set_xlabel("xlabel")
ax.plot(x, y, 'r--')
fig.savefig("fig.png")

Florid

+1 Alternately just use ax = fig.add_subplot(1, 1, 1), or better yet, just fig, ax = plt.subplots(). – Muliebrity

or fig, ax = plt.subplots(1, 1, tight_layout=True) (assuming your version of mpl is new enough) – Victoria

or plt.figure(tight_layout=True) – Foreignism

Didn't work for me trying to create png file, switching to jpg fixed it – Shetrit

Answer 5

I was able to solve the issue (in visual studio code jupyter extension) by changing the format from 'png' to 'jpg', along with the parameter 'plt.subplots(tight_layout=True)'.

Noiseless

Answer 6

I was using Jupyter Notebook and Just change .png to .jpg and my problem is solved now Here is my code:

# changing the size of figure to 2X2
plt.figure(dpi=100, figsize=(15, 10))
plt.grid()
#display(plt.plot(year1, ratio1))
x = np.arange(1900, 2020, 5)
plt.xticks(x)
plt.title(ttile)
plt.xlabel('Year')
plt.ylabel('Ratio')
plt.plot(year,ratio)
plt.savefig('books_read.jpg', dpi = 300)

Saved Image from the Code

Majestic

Helpful! Changing .png to .jpg worked outside Jupyter as well! – Resurrectionist

Answer 7

Comment out plt.show() and instead call plt.savefig('abc.png').

Emancipated