How to use Matplotlib color map in an image

[Gene Beidl]

I am creating a spectrogram to be displayed as an image and I want to use the magma colormap from Matplotlib.

It's shown here:

https://matplotlib.org/users/colormaps.html

And some 'magma' mapped spectrograms are here:

https://librosa.github.io/librosa_gallery/auto_examples/plot_hprss.html

Is there a way to use Matplotlib colormaps with an image in pyqtgraph?

Thanks!

[Patrick]

Hi,

You can set your own custom tick colors and positions on a colormap bar with something like this:

cbar = pg.HistogramLUTItem(image=some_image)
cbar.gradient.restoreState({"mode": "rgb", "ticks":
    [(0.0, (0.374272, 0.119296, 3.549696)),
    (0.125, (28.952064, 16.765952, 70.856704)),
    (0.25, (81.063424, 18.35264, 124.25728)),
    (0.375, (131.284736, 37.933824, 129.957888)),
    (0.5, (183.395072, 55.035392, 121.67424)),
    (0.625, (231.495936, 81.82016, 99.363072)),
    (0.75, (252.5952, 137.108992, 97.84576)),
    (0.875, (255.205888, 197.015296, 136.932352))]})

Those tick positions should be a decent copy of magma, which I created with

from matplotlib import cm
ticks = [(i/256, tuple([256*j for j in cm.magma.colors[i]])) for i in range(0, cm.magma.N, 32)]

Patrick

[Gene Beidl]

 Great, thank you Patrick!

[Erik J]

Hi Gene,

Here is a simple function I use to convert mpl colormaps to pyqtgraph form:

import numpy as np

from matplotlib import cm

def colormap(name):

    colormap = cm.get_cmap(name)

    colormap._init()

    # Convert matplotlib colormap from 0-1 to 0-255 for PyQtGraph

    lut = (colormap._lut * 255).view(np.ndarray)[:colormap.N] 

    return lut

You can easily use it to set your colormap in pyqtgraph as follows:

         img.setLookupTable(lut)

where img is a PyQtGraph ImageItem. The truncation of the colormap ("[:colormap.N]") keeps only the portion with the colors. mpl has a few extra entries at the top of the colormap that don't work with pyqtgraph.  I got the function originally from somewhere in these group postings but I can no longer find it. I added the truncation part. This works well for me as I display both mpl and pyqtgraph images in a single Qt window.

Hope this helps,

Erik

[Gene Beidl]

Great, thank you Eric!