add option for log color scale with px.imshow

[janosh]

To better distinguish similar values, it would be nice if the color scale of px.imshow() could be made logarithmic without changing the values by displayed text_auto=True.

Closest I've found so far is to np.log the df and change the color bar ticks:

df_log = np.log10(df)

fig = px.imshow(
    df_log,
    text_auto=True,
)

ticks = [1, 0.5, 0.2, 0.1, 0.05, 0.02, 0.01]
vals = np.log10(ticks)
fig.update_layout(coloraxis_colorbar=dict(tickvals=vals, ticktext=ticks))

[loehndorf]

I second this idea. I can log-transform the data but then the color scale does not show the right values anymore.

[SengerM]

In case someone ends up here looking for a quick solution, here is my solution. It should be completely compatible with px.imshow, just change to imshow_logscale.

Would be nice to have something like this officially part of Plotly as log scale is very useful.

[gvwilson]

see #3616