[postgis-users] 3d support in postgis

Chris Hodgson chodgson at refractions.net
Tue Oct 15 08:58:47 PDT 2002


The "3d" support you speak of is actually 2.5d support. What this means is
that all of the shapes must be valid 2-d objects, and each point can, in
addition to its x and y coordinates, have a z coordinate. This means that
you cannot represent full 3-d objects (polyhedrons) and that support for
objects whose depth is primarly in the z-axis (eg. a rectangle in the
z-plane ) are not ideally represented.

This also means that indexing only needs to be done on the 2-d bounding box,
and that "truly_inside" only makes sense in 2-d (because we have no way to
represent an object with volume in 2.5d).

The reason for 2.5d is that it is much simpler to implement and full 3d is
really unneccesary for geographical applications. What sort of 3-d data are
you trying to work with in postgis?

Chris

----- Original Message -----
From: "Wolfgang Bär" <WBaer at gmx.de>
To: <postgis-users at postgis.refractions.net>
Sent: Monday, October 14, 2002 9:17 AM
Subject: [postgis-users] 3d support in postgis


> hi everyone,
>
> i have a question regarding the 3d support in postgis.
>
> are 3d object indexed with the complete boundingbox, which means the
> search for overlaps also test for overlapping in the z coordinate
> during index lookup?
>
> in the faq of the postgis manual in question 5 it is said - use the
> truly_inside function for the search for object really inside the
> search box - but this function only tests the x and y coordinate and
> ignores the z coordinate which will result in wrong results when
> dealing with 3d data.
>
> does anybody knows if the gist-index operates with the z coordinate
> and only some functions are not extended to 3d at this time !
>
> thanks for any help,
>
> bye wolfgang
>
>
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users
>
>





More information about the postgis-users mailing list