Matplotlib colourmaps for QGIS

README.md

mpl2qgis

See script below.

Make sure the Python file is executable. Then:

$ ./mpl2qgis.py viridis bone

This writes a file colourmaps.xml. Result:

<!DOCTYPE qgis_style>
<qgis_style version="0">
<symbols/>
<colorramps>
    <colorramp type="gradient" name="viridis">
    <prop k="color1" v="68,1,84,255"/>
    <prop k="color2" v="253,231,37,255"/>
    <prop k="stops" v="0.04;71,16,99,255:0.08;72,29,111,255:0.12;71,42,122,255:0.16;69,55,129,255:0.2;65,68,135,255:0.24;60,79,138
,255:0.28;55,90,140,255:0.32;50,100,142,255:0.36;46,111,142,255:0.4;42,120,142,255:0.44;38,130,142,255:0.48;34,139,141,255:0.52;31,1
49,139,255:0.56;31,159,136,255:0.6;34,168,132,255:0.64;44,177,126,255:0.68;59,187,117,255:0.72;78,195,107,255:0.76;99,203,95,255:0.8
;122,209,81,255:0.84;149,216,64,255:0.88;176,221,47,255:0.92;202,225,31,255:0.96;229,228,25,255"/>
    </colorramp>
    <colorramp type="gradient" name="bone">
    <prop k="color1" v="0,0,0,255"/>
    <prop k="color2" v="255,255,255,255"/>
    <prop k="stops" v="0.04;9,9,12,255:0.08;18,17,24,255:0.12;26,26,37,255:0.16;35,35,49,255:0.2;45,45,62,255:0.24;53,53,74,255:0.
28;62,62,86,255:0.32;71,71,99,255:0.36;80,80,112,255:0.4;89,92,121,255:0.44;98,104,130,255:0.48;107,116,139,255:0.52;116,129,148,255
:0.56;125,141,157,255:0.6;134,154,166,255:0.64;143,166,174,255:0.68;152,179,184,255:0.72;161,191,193,255:0.76;172,202,202,255:0.8;18
5,210,210,255:0.84;200,220,220,255:0.88;214,229,229,255:0.92;228,238,237,255:0.96;241,246,246,255"/>
    </colorramp>
</colorramps>
</qgis_style>

To install the color ramps, go to QGIS > Settings > Style Manager > Color ramp, hit the green + sign, and select the file.

Why

https://twitter.com/AndyDoggerBank/status/1085527532470968321

Inspiration

http://rocksandwater.net/blog/2016/07/qgis_perceptually_uniform_colorramps/

mpl2qgis.py

#!/usr/bin/env python
#-*- coding:utf-8 -*-
"""
Converts matplotlib colourmaps to QGIS colourmaps.
Matt Hall, 2019, released into public domain / CC-0.
"""
import sys

from matplotlib import colormaps as mpc

XML = """<!DOCTYPE qgis_style>
<qgis_style version="0">
  <symbols/>
  <colorramps>
{0}
  </colorramps>
</qgis_style>"""

CRAMP = """    <colorramp type="gradient" name="{name}">
      <prop k="color1" v="{color1}"/>
      <prop k="color2" v="{color2}"/>
      <prop k="stops" v="{stops}"/>
    </colorramp>"""


def get_rgba(name, value):
    """Get (R, G, B, A) for a value on a matplotlib colormap.
    """
    cmap = mpc.get_cmap(name)
    rgba = [str(int(round(255*v, 0))) for v in cmap(value)]
    return rgba    


def qgis_xml(name):
    params = {'name': name}
    params['color1'] = ','.join(get_rgba(name, 0.0))
    params['color2'] = ','.join(get_rgba(name, 1.0))
    stops = []
    for v in range(4, 100, 4):
        v /= 100
        s = f'{v};'
        s += ','.join(get_rgba(name, v))
        stops.append(s)
    params['stops'] = ':'.join(stops)
    return CRAMP.format(**params)


def mpl2qgis(names):
    cramps = '\n'.join(qgis_xml(name) for name in names)
    return XML.format(cramps)


if __name__ == '__main__':
    with open('colourmaps.xml', 'w') as f:
        f.write(mpl2qgis(sys.argv[1:]))

@JesusLopTem You need to be on Python 3.6+ for f-strings. You can change that line to s = '{};'.format(v) and I think it will work.

Ok now. Thanks!

Hello, I was wondering if you know how to achieve the inverse. I mean, how can you use a QgsColorRamp in a Matplotlib graph instead of a Matplotlib colormap?

Thank you for the genuinely nice utility. I would suggest change the import in line 9 to:

from matplotlib import colormaps as mpc

This is because of the following deprecation warning:

MatplotlibDeprecationWarning: The get_cmap function was deprecated in Matplotlib 3.7 and will be removed two minor releases later. Use ``matplotlib.colormaps[name]`` or ``matplotlib.colormaps.get_cmap(obj)`` instead.

I'm using matplotlib 3.7.2 in python 3.11.4.

@jpdarela Thank you! Done.

@AkliDif Assuming you can find the XML file, you would have to parse it into an array and then do something like this

If all you have is an image of the colourmap, then unmap might be able to recover a matplotlib colourmap from it.