http://boston.freemap.in/bugs/2221/poor_follow.png
Uses:
http://boston.freemap.in/bugs/2221/poor-follow.map
All data in:
http://boston.freemap.in/bugs/2221/
Command used:
shp2img -m poor-follow.map -o poor_follow.png -e 232376.94043 900487.441406 232524.42627 900634.927246 -l roads_3_1
Discussion on IRC:
20:24:28 < hobu> Someone needs to do some splining
20:24:29 < crschmidt> Should I bother to file it as a bug?
20:25:06 < hobu> You could file an enhancement for 5.2 that says, 'we should improve ANGLE FOLLOW to support simple interpolation methods like splining for degenerative cases"
20:25:24 < hobu> It's way too big to be dealt with in the throws of our 5.0 release right now though
20:25:40 < crschmidt> understood
20:25:46 < crschmidt> Just wasn't sure if it was worth filing at all
20:26:19 < hobu> Another thing about the splining is we could maybe take advantage of it for feature rendering as well
20:26:38 < hobu> something those paleogeographers do from time to time ;)
<crschmidt> what is that?
<crschmidt> straightening out of too-curvy borders?
<hobu> smoothen disjointed ones. They add map noise for very little information
* crschmidt nods
<crschmidt> something like what a simplification algorithm would do
<crschmidt> only automatic/built into the rendering?
<hobu> right, except not one that has to preserve topological relationships
<hobu> http://en.wikipedia.org/wiki/B-spline
<hobu> Would be a simple approach
<hobu> Maybe you'd tell the mapserver renderer, 'hey, use 6-8 consequtive vertices to calculate your splines, if the incident angle is less than x degrees"
<hobu> or something like that ;)
<hobu> It'd be spendy, for sure
<hobu> and not very useful for Manhattan grid systems :)
