[OSM-dev] Shaded relief in OSM

David Earl david at frankieandshadow.com
Thu Dec 6 14:45:28 GMT 2007


On 06/12/2007 13:51, Artem Pavlenko wrote:
> Yes, this is my feeling as well. The resolution is no good for high  
> zoom levels.
> Maybe we should look into creating our own relief maps, specially  
> tailored to the OSM.

One thought I had was that at high zooms an absolute relief map isn't 
much help. You can have relatively flat high or low areas. If you want a 
feel for what the landscape looks like, you need a relative relief map 
at somewhat higher vertical resolution. But then how do such things join 
up as you pan, and avoid having an area of constant colour that could 
either be the flank of a hill at right angles to the "light source" or a 
local plateau.

I was thinking one could analyse the absolute relief data with a pixel 
filter so that each pixel gets shaded according to a difference function 
between its altitude and that of all the other pixels in the filter (say 
the neighbouring 10 x 10 grid). You'd end up with a vector whose 
direction gives the direction of fall at the point, and whose length 
indicates how extreme the gradient is. If say one used for a shade of 
green and the other combined with the light source direction to give a 
shade of gray you might just get a useful result which shows the local 
terrain independent of absolute altitude.

Of course, we have none of these issues in Cambridge because it is flat 
:-) Even Castle Hill might only just show up in a relative relief map.

David




More information about the dev mailing list