Due to popular request, I've modified the d.legend range= option to
extend to the limits given in the colr file, if it exists.
This will allow a user to copy a standard colr file to many rasters and
have the same legend display regardless of the actual data extent. Some
remote sensing applications where this might be useful are for frames of
a SST animation or display of 0-255 ranged color channel data.
While this works pretty well, I'm still not 100% convinced it's a good
idea. My main problem with it is conceptual: that d.legend isn't purely
a child of the raster data anymore.
If somebody wants to mis-represent a what's in a plot, I think they
should have to do it explicitly with the legend taken from a second
raster, and not by oversight. I've put in a Warning message, but I'm
still not totally happy about it.
My second problem is I'm not sure how it will interact with the
still-not-committed use= option, which takes precedence. It requires
some think, but probably will be ok.
anyway, here it is. The diff is versus today's CVS HEAD.
Due to popular request, I've modified the d.legend range= option to
extend to the limits given in the colr file, if it exists.
This will allow a user to copy a standard colr file to many rasters and
have the same legend display regardless of the actual data extent. Some
remote sensing applications where this might be useful are for frames of
a SST animation or display of 0-255 ranged color channel data.
I hope that I understand this option correctly, if yes then
another useful application is time series of data either from
monitoring or simulations, where e.g. for water I use the same color
table
for all time steps although at the beginning of the simulation the range
of data
is much smaller. It is even more useful when comparing results of
simulations
(e.g. erosion) for different scenarios, by using the same color table
and same legend for all results you quickly see which one produce much
less
erosion than the maximum.
So I consider this option really useful and I hope that it does not
broke anything
and does not cause some unexpected problems.
thank you for working on legends - it has made a whole world of
difference for my work,
Helena
While this works pretty well, I'm still not 100% convinced it's a good
idea. My main problem with it is conceptual: that d.legend isn't purely
a child of the raster data anymore.
If somebody wants to mis-represent a what's in a plot, I think they
should have to do it explicitly with the legend taken from a second
raster, and not by oversight. I've put in a Warning message, but I'm
still not totally happy about it.
My second problem is I'm not sure how it will interact with the
still-not-committed use= option, which takes precedence. It requires
some think, but probably will be ok.
anyway, here it is. The diff is versus today's CVS HEAD.