[OSM-dev] Free mapping software

[Richard Fairhurst]

Lars wrote:

> bvh wrote:
>
>> It seems there has been talk about using bezier segments.
>> What's the status on that?
>
> Many of us here are new to GIS, but have a general computing and
> graphics background.  It is natural to think of bezier curves, but
> it appears nobody in GIS use them, and probably for a good reason.

Mmm, maybe.

"Industry standard" GIS is an enormously conservative science where  
it's traditionally assumed you need a heavyweight package like ArcGIS,  
MapInfo or whatever to do even the most simple spatial analysis. The  
GIS industry is having a devil of a time keeping up with today's more  
agile development ethos (Google Earth is the most prominent example):  
indeed, OSM also hasn't thus far taken much from GIS, and that's  
probably all to the good.

You're right, nobody in GIS uses beziers, but that might be one of the  
reasons so many generated-from-GIS maps look dog awful. Even on OSM  
it's detectable - try zooming into a roundabout and look how angular  
the paths are; unless you drop 20 points around even the smallest  
roundabout, there's a clear difference from the situation on the  
ground. _Cartographers_ do often use beziers and the difference shows.

At OSM we should be aiming for the best possible output; yes, we're a  
long way off now, but we've come a long way in two years. FWIW I don't  
really think that beziers are a priority right now, but I can see a  
day when they start to become a part of the data model, maybe allied  
with an "optimise for delivery" approach where certain clients store  
straight-line approximations at small scales. (Projection is less  
significant at large scales, making it simpler to render straight from  
the bezier.)

cheers
Richard