homogenous numpy array renders same matplotlib cmap regardless of value

Question

I was trying to debug some visualization code and when I worked back to ultra-basics, I discovered a confusing behavior that makes me think I'm missing something fundamentally about matplotlib cmaps.

This works as i expect -- printing a range of reds and blues corresponding to randomly generated values between 0,1.

fig = plt.figure()
CMAP =  "seismic"   # blue-white-red colormap
a = np.array(np.random.rand(50,50)/2,).reshape(50,50)
plt.imshow(a, cmap=plt.get_cmap(CMAP))

But if I make the array values homogenous, the plot is always the same dark blue value regardless of what the array's value is.

fig = plt.figure()
CMAP =  "seismic"   # blue-white-red colormap

#either of these two lines gives the same result
a = np.array([0.9]*50*50).reshape(50,50)
a = np.array([0.1]*50*50).reshape(50,50)

As long as the array doesn't have homogenous values, it seems to work as expected, for example

a = np.array([0.1,.9]*25*50).reshape(50,50)

renders with stripes, as I'd guess.

JohanC · Accepted Answer · 2020-08-01 09:17:21Z

2

To see what is happening, it is helpful to add a colorbar next to the image:

import numpy as np
import matplotlib.pyplot as plt

cmap =  "seismic"
a = np.random.rand(50,50)/2
plt.imshow(a, cmap=cmap)
plt.colorbar()
plt.show()

Note that dark blue corresponds to the value 0 and dark red to 0.5.

The following shows what happens in the other cases:

fig, (ax1, ax2, ax3) = plt.subplots(ncols=3, figsize=(12, 3))
cmap = "seismic"

a = np.full((50,50), 0.9)
img1 = ax1.imshow(a, cmap=cmap)
plt.colorbar(img1, ax=ax1)

a = np.full((50,50), 0.1)
img2 = ax2.imshow(a, cmap=cmap)
plt.colorbar(img2, ax=ax2)

a = np.repeat([0.1,.9], 25*50).reshape(50,50)
img3 = ax3.imshow(a, cmap=cmap)
plt.colorbar(img3, ax=ax3)

When all data values are 0.9, matplotlib chooses a small range (in this case from 0.825 to 0.975) where the constant value 0.9 now corresponds to white. (Your example code seems to use an older version of matplotlib where the constant value corresponds to the darkest color). Something similar happens for the constant 0.1, with another range.

With half 0.1 and half 0.9, the range is set from 0.1 to 0.9, with 0.1 corresponding to dark blue and 0.9 to dark red.

These values are accessible via parameters, they are called vmin and vmax. When drawing many plots it can be interesting to use the same vmin and vmax for all of them. Here are the same examples, but with imshow(..., vmin=0, vmax=1):

answered Aug 1, 2020 at 9:17

JohanC

81.4k8 gold badges54 silver badges90 bronze badges

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

1 Comment

user3556757 Over a year ago

hey thanks for this excellent answer. IT answered my question AND enabled me to solve the trickier deeper problem. Basically the matplotlib axis was rescaling in ways I didn't expect. vmin, vmax fixed it.

Collectives™ on Stack Overflow

homogenous numpy array renders same matplotlib cmap regardless of value

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related