Support for custom colormaps

[lopezvoliver]

This pull request adds support for custom colormaps (resolves #200), either as a matplotlib Colormap, or as a list of colors.

Hopefully I got it right, so far my tests (with leafmap.add_raster) seem to work:

Here's a test using the example from opengeos/leafmap#673

[giswqs]

This would be a nice addition. Great work!

[banesullivan]

Fantastic! I'll give this a test tonight and see if I can write a few tests. Thank you so much for contributing this!

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

Attention: Patch coverage is 77.77778% with 6 lines in your changes missing coverage. Please review.

Project coverage is 83.74%. Comparing base (33838b3) to head (04f4cd5).

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #223      +/-   ##
==========================================
- Coverage   83.93%   83.74%   -0.19%     
==========================================
  Files          26       26              
  Lines        1021     1046      +25     
==========================================
+ Hits          857      876      +19     
- Misses        164      170       +6     

[banesullivan]

This isn't quite working as expected for colormaps. I'm eager to incorporate this but we'll have to iterate on the serialization of the colormaps to account for different types

[lopezvoliver]

This isn't quite working as expected for colormaps. I'm eager to incorporate this but we'll have to iterate on the serialization of the colormaps to account for different types

Can you share a couple of examples? I can give it a try

[banesullivan]

Sorry for my short comment, I'm unable to take a look immediately, but I was trying to do something like the following:

from localtileserver import TileClient, get_leaflet_tile_layer, examples
from matplotlib.colors import ListedColormap, Colormap
from ipyleaflet import Map

tile_client = examples.get_bahamas(debug=True)

cmap1 = ListedColormap(['red','blue'],name='test_1')
cmap2 = ListedColormap(['blue','green'],name='test_2')
cmap3 = ListedColormap(['green','red'],name='test_3')

t = get_leaflet_tile_layer(tile_client, colormap=cmap1)
# t = get_leaflet_tile_layer(tile_client, colormap=[cmap2,cmap1,cmap3])

m = Map(center=tile_client.center(), zoom=tile_client.default_zoom)
m.add_layer(t)
m

[lopezvoliver]

@banesullivan can you explain a bit more what's supposed to happen with colormap in this line? is it going to be a concatenation of the three colormaps?

t = get_leaflet_tile_layer(tile_client, colormap=[cmap2,cmap1,cmap3])

What about the example with the single colormap? is that one working as expected?

[banesullivan]

can you explain a bit more what's supposed to happen with colormap in this line? is it going to be a concatenation of the three colormaps?

Sorry for my delay! Let's ignore this actually. this was a supported feature when the underlying tile server used to be large-image where it had some rather unique capabilities for color mapping individual bands. I don't think we need to support that here.

[banesullivan]

I've gone ahead and fixed the remaining issues in order to use ListedColormap and LinearSegmentedColormap instances as well. I think this is good to merge at this point. @lopezvoliver or @giswqs, do you want to make sure that this works with leafmap before I merge/release?

[banesullivan]

I think its worth noting that in this current implementation, there is a limit to how big these colormaps can be as we embed them in the URL parameters. I recommend just creating named colormaps and registering them with matplotlib in your environment to get around this so that they can be referred to by name, ref: https://matplotlib.org/stable/users/explain/colors/colormap-manipulation.html#registering-a-colormap

[giswqs]

I tested it. Custom colormaps in the format of ListedColormap and list of colors (e.g., ['red', 'green']) work well with leafmap.

Registering the custom colormap with matplotlib does not seem to work. Anyway, this is not a big issue. No worries if you don't have time look into it now.

https://colab.research.google.com/drive/1JPgl3O-u1ufDEwb3f8fWxiXPqOQyPER7?usp=sharing

[banesullivan]

Registering the custom colormap with matplotlib does not seem to work

Whoops, I thought this would. I'll try to make sure this gets fixed

[nve-sak]

FWIW, I added a failing test case here: opengeos/leafmap#966. No crash or error, but the colors are not as expected.