How to use viridis in matplotlib 1.4

Question

I want to use the colormap "viridis" (http://bids.github.io/colormap/), and I won't be updating to the development version 1.5 quite yet. Thus, I have downloaded colormaps.py from https://github.com/BIDS/colormap. Unfortunately, I'm not able to make it work. This is what I do:

import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import numpy as np

import colormaps as cmaps

img=mpimg.imread('stinkbug.png')
lum_img = np.flipud(img[:,:,0])

plt.set_cmap(cmaps.viridis)
imgplot = plt.pcolormesh(lum_img)

This gives me a ValueError, the traceback ending with,

ValueError: Colormap viridis is not recognized. Possible values are: Spectral, summer, coolwarm, ...

(And then the complete list of originally installed colormaps.)

Any thoughts on how to fix this issue?

Solution 1

To set viridis as your colormap using set_cmap, you must register it first:

import colormaps as cmaps
plt.register_cmap(name='viridis', cmap=cmaps.viridis)
plt.set_cmap(cmaps.viridis)

img=mpimg.imread('stinkbug.png')
lum_img = np.flipud(img[:,:,0])
imgplot = plt.pcolormesh(lum_img)

Solution 2

Rather than using set_cmap, which requires a matplotlib.colors.Colormap instance, you can set the cmap directly in the pcolormesh call

(cmaps.viridis is a matplotlib.colors.ListedColormap)

import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import numpy as np

import colormaps as cmaps

img=mpimg.imread('stinkbug.png')
lum_img = np.flipud(img[:,:,0])

imgplot = plt.pcolormesh(lum_img, cmap=cmaps.viridis)

Solution 3

What I did is to just copy the

_viridis_data = [[0.267004, 0.004874, 0.329415],
                 [0.268510, 0.009605, 0.335427],
                 [0.269944, 0.014625, 0.341379],
                 :
                 [0.983868, 0.904867, 0.136897],
                 [0.993248, 0.906157, 0.143936]]

from https://github.com/BIDS/colormap/blob/master/colormaps.py

and add:

from matplotlib.colors import ListedColormap

viridis = ListedColormap(_viridis_data, name='viridis')

plt.register_cmap(name='viridis', cmap=viridis)
plt.set_cmap(viridis)

Solution 4

Download the colormaps.py from here,then:

import os,sys
scriptpath = "/Your downloading path/colormap-master/"
sys.path.append(os.path.abspath(scriptpath))
import colormaps as cmaps   

Done!

View more solutions

Comments

[ukrutt]

Awesome, just what I need.

[AnnanFay]

What would cause ImportError: No module named colormaps ?

[tmdavison]

@Annan: Have you downloaded colormaps.py from github.com/BIDS/colormap?

[AnnanFay]

@tom Thanks for replying to my query. I didn't realise it was a separate module (I've been jumping around a few questions about the 'viridis' cmap and did not fully read the question you were answering). I actually fixed the issue by updating matplotlib to the latest version - which for me is 1.5.3.

[tmdavison]

@ Annan. Yes, obviously the best way to get viridis is to use the latest version of matplotlib (i.e. any version >= 1.5). However, this question and answer were about using viridis with a version of matplotlib (1.4) which did not include the viridis colormap