[OSRS-PROJ] Orthographic Projections and MapServer

[Bart Adriaanse]

Gerald,

No flames, just comments, and wishfull thinking...

Being a typical user who knows what a projection is and how to work with
them, but prefers to use a tool like Proj or the GIS at hand rather then
even attempt to do my own this is mostly wishfull thinking i suppose.

I can see clearly now there seems to be no simple solution, you opened my
eyes to realize expecting Proj to deal with this is not realistic, i mean
how would it anticipate on me drawing a polygon or polyline, or maybe even a
transparent globe where points beyond the horizon show through half
transparently ;-)

I guess you're right...

So my comment would boil down to simply the need to be able to find out
which points of a polyline processed by pj_forward were over the horizon,
from what i understand by now i could process the points one by one and
assess the return code, right ?

Best regards,

Bart Adriaanse
www.demis.nl

Original Message ----- 
From: "Gerald I. Evenden" <gerald.evenden at verizon.net>
To: "[OSRS-PROJ]" <osrs-proj at remotesensing.org>
Sent: Thursday, August 21, 2003 9:32 PM
Subject: Re: [OSRS-PROJ] Orthographic Projections and MapServer


> This problem is a purely a graphical problem and hence not a fault
> nor limitation of cartographic projection software other than to
> indicate what points cannot be projected due to invisibility.
>
> The graphical problem is much the same a the typical window or clipping
> problem except that the data is defining the window and not a separate
> set of information such as the limits of the map display---typically
> a rectangular region handled by well known software.
>
> In this case, the border of visibility is a curve which is a circle
> in the spherical earth model.  Note: using the elliptical model makes
> no sense for global presentation scales.  The astronomical term "limb"
> applies to this edge of the visible part of the projection.
>
> The first part of the problem is finding the point where a line segment
> intersects the limb---where the last points is visible and the next is
> not.  This problem is fairly easy by recursively dividing the line into
> halves and adjust the first or last point based upon the visibility of
> the center point.  This is what I do for my coastline maps in my
> documentation.  But since I do not fill I quit the problem at this
> point.
>
> What to do to fill?  Basically, the problem is to join the point
> where the line went beyond the limb and connect a circle with the
> limb's radius to the point where it reentered the limb.  Ah, but which
> way?  If the fill area were always to the left of the line segments, the
> arc would be counter clockwise.
>
> The limb cut is easy and the more I think about it, the limb edge is not
> too bad if the right-/left-handedness of the fill area is known.
>
> Ah yes, there is the problem when all points are outside the limb.
> This is a problem with clipping fill where it needs to be
> determined if the clipping area is totally inside or outside the window.
> I belive this problem also exists here but I have not mentally imaged
> how the area outline would look.
>
> Any comments, flames, ... ?
>
> -- 
> Gerald I. Evenden <gerald.evenden at verizon.net>
>
> ----------------------------------------
> PROJ.4 Discussion List
> See http://www.remotesensing.org/proj for subscription, unsubscription
> and other information.

----------------------------------------
PROJ.4 Discussion List
See http://www.remotesensing.org/proj for subscription, unsubscription
and other information.