[Matplotlib-users] strange behaviour or bad coding? color indices in colorbar and data doesnt match.

[Yasin S. B. <yas...@gm...>]

Hi.

i wanted to make a colorbar for a dataset. dataset has 50 different values.
so instead of entering a different color for 50 values one by one, i tried
the code below.


here is what happens:

1- two different data value is painted same color.

changing the color number to some value, sometimes works, sometimes doesnt.

for example, if i change the N (number of colors) in the code below, from
50 to 52,
problem continues, but with different indices. (29-30) are same instead of
(28-29).

but if i change it to 60 colors, problem "seems to" go away for given data
values in the example.

2- colors in image are not "exactly" taken from colorbar. they are
different a tiny bit.
if you check that darkblue color color picker, there is no such color in
colorbar!

second 'data' line (RGB 21-1-167)
the most close one is color index 28 (RGB 22-2-168)

and even correctly indexed data colors are off a little bit (RGB-values, i
mean).


PS: since colors are very similar and look same, you need to use
a RGB color picker/finder app. (i use Color Selector 1.0)

thanks in advance.

#
===========================================================================

from matplotlib import colors, colorbar, pyplot
import numpy as np

# make a dict for colors.
color_info = {}

color_info["red"] = [[0.0, 0.00, 0.00],  # black
                     [0.2, 0.40, 0.40],  # brown
                     [0.4, 0.60, 0.60],  # purple
                     [0.6, 0.00, 0.00],  # darkblue
                     [0.7, 0.00, 0.00],  # green
                     [0.8, 1.00, 1.00],  # yellow
                     [0.9, 1.00, 1.00],  # red
                     [1.0, 1.00, 1.00]]  # white

color_info["green"] = [[0.0, 0.00, 0.00],
                       [0.2, 0.20, 0.20],
                       [0.4, 0.05, 0.05],
                       [0.6, 0.00, 0.00],
                       [0.7, 1.00, 1.00],
                       [0.8, 1.00, 1.00],
                       [0.9, 0.00, 0.00],
                       [1.0, 1.00, 1.00]]

color_info["blue"] = [[0.0, 0.00, 0.00],
                      [0.2, 0.00, 0.00],
                      [0.4, 0.70, 0.70],
                      [0.6, 0.65, 0.65],
                      [0.7, 0.00, 0.00],
                      [0.8, 0.00, 0.00],
                      [0.9, 0.18, 0.18],
                      [1.0, 1.00, 1.00]]

# divide colorbar into 50 colors. you can change this (Ex: N=15 and use
data2)
my_cmap = colors.LinearSegmentedColormap('my_colormap', color_info, N=50)

N = my_cmap.N

# we put tickmarks into the middle of colors, by shifting half a segment.
ticklabels = np.arange(N)
ind = np.arange(N)
ticks = [(i / float(N)) + (1 / float(2 * N)) for i in ind]


def my_cbar(cbar_ax, cmap=my_cmap):
    """adds cbar to image."""
    base = colorbar.ColorbarBase(cbar_ax, cmap=my_cmap,
                                 ticks=ticks,
                                 orientation="horizontal")

    xlabels = cbar_ax.get_xticklabels()
    for label in xlabels:
        label.set_rotation(60)

    base.set_ticklabels(ticklabels)

# labels are very close so dont forget to maximize plot window to see
better.
fig = pyplot.figure(figsize=(7, 7))

data = [[27, 28],
        [28, 29],  # both are same colors!
        [29, 30],
        [30, 31],
        [31, 32],
        [32, 33]]
# data2 is just to show it works as expected for N=15.
data2 = [[7.999, 8],
         [8.999, 9],
         [9.999, 10],
         [10, 11],
         [11, 12],
         [12, 13]]

# with these vmin/vmax, data value 0-0.99999 goes to 'color index 0',
# data value X - X.9999 goes to 'color index X' ...etc
# (change 'data' to 'data2' when N=15)
pyplot.imshow(data, cmap=my_cmap,
              vmin=0, vmax=N, interpolation="nearest")


cx = fig.add_axes((0.2, 0.05, 0.6, 0.03))
my_cbar(cx, cmap=my_cmap)


# but data values 28 and 29 goes to same color. maybe this happens for other
# data values as well ???

# similarly interesting, if you check that darkblue color with a
# rgb color picker/reader, there is no such color in colorbar! the most
close
# one is color index 28.
# RGB values ->>> 'data[1, :]=(21, 1, 167)' vs 'color index 28=(22, 2, 168)'

# and even correctly indexed colors RGB values are off a little bit.
pyplot.show()

-- 
Yasin Selçuk Berber
"Bismillah, her hayrın başıdır."