[Proj] Problem understanding the new API

[Idan Miara]

Thanks.
I appreciate your detailed answer!

On 20 Feb 2018 08:46, "Kristian Evers" <kreve at sdfe.dk> wrote:

>
>
> > On 20 Feb 2018, at 06:21, Idan Miara <idan at miara.com> wrote:
> >
> > Hi,
> >
> > Which CRS is exactly the hub in the old API? Is it WGS84 lat/lon or
> geocentric?
>
> WGS84 lat/lon, although if necessary the coordinates are passed through a
> geocentric WGS84 (when using +towgs84).
>
> > Should it be more accurate or faster to use the new API to transform
> WGS84 lat/lon to WGS84 UTM,  will it use any different transformation path?
> >
>
> Well… It can be more accurate if you know which WGS84 realisation you are
> working in,  you have timestamps for your data AND you define your own
> custom transformation. Otherwise the accuracy will be the same. If you
> don’t know anything about your data then assume that the uncertainty of the
> transformation is 2m. If it is faster will depend on what you are doing.
> Initial testing show that the old and new APIs perform more or less the
> same for similar transformations. In general transformations will probably
> be slightly slower since more complicated transformations can be done now.
>
> You CAN use a different transformation path if you want - and know how to
> - otherwise no, the transformation path will be the same for now, for
> example if you are using the epss-init file with the proj_create_crs_to_crs
> function. At the moment using a different transformation path requires you
> to set up a transformation pipeline that does what you want. In the future
> we will improve the situation such that at least the epsg transformations
> uses the optimal transformation path in a given situation.
>
> > Thanks.
> >
>
> _______________________________________________
> Proj mailing list
> Proj at lists.maptools.org
> http://lists.maptools.org/mailman/listinfo/proj
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20180220/d4e076d5/attachment.html>