Ævar Arnfjörð Bjarmason a scris: > 2009/6/1 Eddy Petrișor <[email protected]>: >> Use a logarithmic scale for your "heat" input (probably is best to use >> 10 as the base). So instead of min use log(min), instead of max, use >> log(max) and so on. > > I tried generating a map with log($size)/log($max): > > http://u.nix.is/~avar/osm-heatmap-log.png > > I tried playing around with it a bit more, e.g. by using log10. But > mostly I get maps that are too hot. Or was log($size)/log($max) not > what you meant?
I was aiming more towards what Robert said and looking at the scale of your values (avg is 2 orders of magnitude higher than min, max is 2 orders of magnitude higher than avg) I expect an improved result when looking at the normalized (relative to log(min) result). OTOH I am a pinch worried that it might be too cold on the high part of teh spectrum, since while min and avg are around the half of their orders of magnitude (both are start with 5), the max is a little lower in the high 3-4 zone. If that is the case (you find the heat zones are not hot enough) you could try to punch up the colours by adding some constant to the logarithm based function of the input: e.g. (log($size) - log($min) + 0.5) / (log($max) - log($min)) which for your numbers could be approximated to: (log($size) - 3.23) / log ( $max / $min ) aproximated further to: log($size)/3.85 - 0.84 Which is quite simple to calculate. Later play with the offset (0.84) to adjust the heat levels and see what works best. > $size / ($median * 32) works pretty well though: > > http://u.nix.is/~avar/osm-heatmap-black-2.png > -- Regards, EddyP ============================================= "Imagination is more important than knowledge" A.Einstein
signature.asc
Description: OpenPGP digital signature
_______________________________________________ dev mailing list [email protected] http://lists.openstreetmap.org/listinfo/dev
