[OSRS-PROJ] Orthographic Projections and MapServer
B at rt.nl
Thu Aug 21 14:30:52 EDT 2003
> The only sure method I know of is to clip the polygon to the great circle
> is the 'rim' of the projection prior to submitting it to the draw routine
> perhaps also the proj forward projection.
> > One solution would be to somehow detect the points which are beyond the
> > horizon and project those to lie on the horizon instead, this would
> > potentially render some polygon border lines at the horizon which
> > really be there, but it seems like a good compromise.
I suppose we agree on this, but would you agree it would be a nice feature
of the projection engine to have this sensible solution built in by default
> > However, even befor thinking about how to reproject them to the horizon,
> > had problems detecting which points were beyond the horizon to begin
> For the spherical orthographic case, which at the scale where the Earth's
> is visible is the most sensible, if a point is more then 90* from the
> the projection: it is beyond the horizon. This is probably easiest to do
> direction cosine form < X,Y,Z unit vectors instead of Lat Lon > of the
When i ran into this problem i think i was using another projection, the
fact that many projections will encounter the horizon problem (sometimes in
extreme cases only) pleads for having the projection engine deal with
instead of expecting the calling application to solve such a basic problem.
I'm affraid there's little more i can contribute at this point, just
wishfull thinking... ;-)
PROJ.4 Discussion List
See http://www.remotesensing.org/proj for subscription, unsubscription
and other information.
More information about the Proj