Using matplotlib, how can I print something "actual size"?

Question

I have a number of plots where the x- and y-axes are in centimeter units, and I am already using axis('equal') to ensure proper aspect ratios. I would like to print out those plots so that when I measure distances within my axes, the distances correspond to real-world centimeters.

That is, a line that is 3 units (cm) long in the plot should print out to be 3 cm long. (A more complex example would be drawing a ruler in Matplotlib, and then printing it out for use /as/ a ruler.) I found solutions in matlab and mathematica, but not for Matplotlib. Is there a magic formula to achieve this? I believe it will require a special combination/ordering of: figure(figsize=??), axis('equal'), fig.canvas.draw(), fig.savefig('filename',format="??"), possibly some math with fig.bbox parameters, and perhaps one or more dpi-settings. I have tried many combinations but haven't hit on the right one. And maybe there is a way-easier approach ...

I think the missing step is forcing the Axes object to be a known proportion of the measured figure size, and then setting the axis limits so that the data-units match the real-world units. — cphlewis
– cphlewis, Commented Apr 1, 2015 at 22:59

Pablo Reyes · Accepted Answer · 2015-04-01 23:30:38Z

4

Consider this example. Where I specify exactly the dimension of my axes in cm. matplotlib works in inches, so I convert to inches. And then I also save it with a particular dpi (128) so that it matches the designed dimensions in my display. This of course varies for every display. I found that by trial and error, even though there might be other methods. Well here the code:

left_margin = 1.   # cm
right_margin = 1.  # cm
figure_width = 10. # cm
figure_height = 7. # cm
top_margin = 1.    # cm
bottom_margin = 1. # cm

box_width = left_margin + figure_width + right_margin   # cm
box_height = top_margin + figure_height + bottom_margin # cm

cm2inch = 1/2.54 # inch per cm

# specifying the width and the height of the box in inches
fig = figure(figsize=(box_width*cm2inch,box_height*cm2inch))
ax = fig.add_subplot(111)
ax.plot([1,2,3])

fig.subplots_adjust(left   = left_margin / box_width,
                    bottom = bottom_margin / box_height,
                    right  = 1. - right_margin / box_width,
                    top    = 1. - top_margin   / box_height,
                    )
fig.savefig('ten_x_seven_cm.png', dpi=128)
# dpi = 128 is what works in my display for matching the designed dimensions.

answered Apr 1, 2015 at 23:30

Pablo Reyes

3,1431 gold badge23 silver badges31 bronze badges

Sign up to request clarification or add additional context in comments.

7 Comments

cphlewis Over a year ago

This is in the display -- does it work with printers? (the original question)

Pablo Reyes Over a year ago

If you open the saved figure and print it in 100% scale, it should match the specified size. For the printer, the dpi will only matter for the resolution of the image.

cphlewis Over a year ago

I just tried that and it didn't; much bigger than scale. Not even isometric -- a y-unit is 3cm on paper, an x-unit is 5cm. This is printed 100% scale, no auto-anything, View Actual Size in the viewer, everything I could think of. (I believe printers are mostly unreliable at this; it's plotters that are accurate.)

Pablo Reyes Over a year ago

What are you using for viewing/printing? I used the mac OS X preview. Then in the printer menu, The Scale is 100% and it did the job. Saving as pdf also reflected the size selected.

cphlewis Over a year ago

OS X Preview, standard desktop printer. Will try GIMP.

|

cphlewis · Accepted Answer · 2015-04-02 00:36:30Z

4

Add this to @pablo reyes' answer, check that the printer is at 100%, and it's pretty close;

ax.set_ylim(0,7)
ax.set_xlim(0,10)

ax.plot([0.5, 1.5],[0.25, 0.25],label='One cm?')
ax.plot([6,6],[1,2], label='One cm?')
ax.legend()

we force the axis to be a size we know, we make its data-transform match the real world, and we can "print a ruler".

answered Apr 2, 2015 at 0:36

cphlewis

16.3k4 gold badges52 silver badges58 bronze badges

5 Comments

Pablo Reyes Over a year ago

Now I understand the confusion. I was more concern about the real dimension of the axes.

cphlewis Over a year ago

Could one write a transReal transform that gets as close as the computer can to "print a ruler"? t I think the Axes would have to get information from the Figure that the Figure doesn't normally pass on, so, maybe not, and certainly a pain.

GaryS Over a year ago

Thank you all! Pablo's solution mostly worked, after modifying dpi to 72. In adding cphlewis' piece, I ran into some bug (oddity?): my axis([xmin, xmax, ymin, ymax]) command is ignored ... same problem with set_xlim() and set_ylim() ... maybe a weird interaction with axis('equal')? The workaround was to plot everything, let matplotlib pick the x/y limits, try to force them (unsuccessfully), use ax.get_xlim() and ax.get_ylim() to determine what the actual limits are (not what I set them to), calc figure_width and figure_height from that, and continue with Pablo's solution. Thanks again!

cphlewis Over a year ago

I think axis([xo, x1, y0, y1]) is in terms of proportion of the Figure, not the data transform. Did you try ax.set_xlim, etc., instead of defining the axis size originally?

cphlewis Over a year ago

also, you won't need axis('equal') if you define them to be equal when printed -- so I'd take it out, to skip the workaround.

Dave X · Accepted Answer · 2025-11-21 19:00:42Z

Another method using fig.add_axes to set the data limits to match the physical plot was quite accurate. I have included 1 cm grid as well.

import matplotlib.pyplot as plt
import matplotlib as mpl

# This example fits a4 paper with 5mm margin printers

# figure settings ( cm size of charted axes plus margin for tick lables) 
figure_width = 28.7 # cm 
figure_height = 20 # cm
left_right_margin = 1 # cm
top_bottom_margin = 1 # cm

# Don't change
left   = left_right_margin / figure_width # Proportion of width
bottom = top_bottom_margin / figure_height # Proportion of height
width  = 1 - left*2 
height = 1 - bottom*2
cm2inch = 1/2.54 # inch per cm

# specifying the width and the height of the box in inches
fig = plt.figure(figsize=(figure_width*cm2inch,figure_height*cm2inch))
ax = fig.add_axes((left, bottom, width, height)) # proportion of figure

# limits settings (important) 
plt.xlim(0, figure_width * width) # width of limits in data units
plt.ylim(0, figure_height * height) # height of limits in data units

# Ticks settings
ax.xaxis.set_major_locator(mpl.ticker.MultipleLocator(5))
ax.xaxis.set_minor_locator(mpl.ticker.MultipleLocator(1))
ax.yaxis.set_major_locator(mpl.ticker.MultipleLocator(5))
ax.yaxis.set_minor_locator(mpl.ticker.MultipleLocator(1))

# Grid settings
ax.grid(color="gray", which="both", linestyle=':', linewidth=0.5)

# your Plot (consider above limits)
ax.plot([1,2,3,5,6,7,8,9,10,12,13,14,15,17])

# save figure ( printing png file had better resolution, pdf was lighter and better on screen)
plt.show()
fig.savefig('A4_grid_cm.png', dpi=1000)
fig.savefig('tA4_grid_cm.pdf')

Results:

Collectives™ on Stack Overflow

Using matplotlib, how can I print something "actual size"?

3 Answers 3

7 Comments

5 Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

7 Comments

5 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related