Dynamic default colormaps

[andreas-wuestefeld]

patch.viz.waterfall() is a powerful tool, creating beautiful figures.
the default "bwr" colormap is great for plotting waveforms (although 'RdBu' is visually more pleasing to me )
Viridis is standard, but also overused.

However, when looking at transformed data, it may be better to have dedicated default colormaps for each transform.

This can be handled in the waterfall function itself, using the "data_type" attribute of the patch. Something like

if "data_type" == 'strain_rate":
    cmap = 'RdBu_r'
if "data_type" == 'strain":
    cmap = 'seismic'
elif  "data_type" == 'FBE":
    cmap = 'Spectral_r'
elif  "data_type" == 'stalta":
    cmap = 'RdGy_r'
elif  "data_type" == 'spectrogam":
    cmap = 'inferno'
else:
    cmap = 'bwr'

This can be handled more efficiently in a map, but this was just to illustrate.

One advantage I see is that it is clear already from the colormap what type of data you are looking at; no need to read the caption or labels. There is a tendency to use viridis for everything, and that is not intuitive.

Any thoughts if this would be a desirable behavior?

[d-chambers]

Yes, i like the idea. Using a data_type color map sounds good, but it should only trigger if cmap was not explicitly defined. That means changing the default value to None, then triggering the dict lookup when cmap is None. I don't really have a strong opinion about defaults, so feel free to adjust as you see fit.