How can I make a logarithmic color plot and colorbar

When working in Matlab 2014a I had a work-around to make logarithmic colorbars for my pcolor plots, making figures such as below.

After an institutional change to Matlab 2018a this workaround is completely broken, and it seems logarithmic colorbars with minor tickmarks are still not natively supported despite being a common type of plot. Is there any method to create this type of plot and colorbar in the newer releases?

Answer 1

[Jeffrey Lestz]

I have version 2018a, and the following also works for me:

set(gca,'colorscale','log')

[Héctor Carrasco López]

That worked for me too. Thank you so much!

[Walter Roberson]

This was introduced in R2018a.

[nlm]

How to get logarithmic scale colors on graph but linear colorbar ?

Answer 2

[Walter Roberson]

cb = colorbar();
cb.Ruler.Scale = 'log';
cb.Ruler.MinorTick = 'on';

[Robin Ramstad]

Hi Walter, thank you for replying. Unfortunately the solution you provided gives me a warning and no ticks or labels at all.

Warning: Error updating ColorBar.

   DataSpace or ColorSpace transform method failed

[Walter Roberson]

Which version are you using? It worked on a small test graph I did in R2018a.

[Antoine]

Hi, I am trying to build the same kind of plots (pcolor with log ZScale). I tried your suggestion on Matlab R2017b and I have the same warning. The colorbar does display some log-scale ticks… but the labels displayed do not correspond at all to the value being plotted. Is there something to adjust after changing this setting? Or do you need to manually apply some transformation to the colormap?

I also tried set(gca, 'ZScale', 'log') and in this case I get:

Warning: Error updating Contour.

   DataSpace or ColorSpace transform method failed

For reference, it looks like this script used to do the trick with previous matlab versions: https://fr.mathworks.com/matlabcentral/fileexchange/29317-logzplot, but it does not seem to work anymore.

I am really surprised that this is still not a default option in Matlab…

[Walter Roberson]

Ah, I do replicate the message in R2017b, but not in R2018a.

I do not see any public bug report about this.

[Jonah Pearl]

Thank you, this works! @Antoine, your error is because you have some data equal to 0 in the Z-direction, and since log(0) is -Inf, the transform fails. Adding a very small number like 0.0001 to your Z-data should fix this. Or you could add in an if statement, if data == 0, add 0.0001.

[Walter Roberson]

Jonah, the error message Antoine shows depends on version . It was a limitation between r2014b and r2017b that did not depend upon data being 0.