[gdal-dev] GeoTransform confusion

[Even Rouault]

Le jeudi 19 novembre 2015 10:47:50, Simon Lyngby Kokkendorff a écrit :
> I can confirm also that it works when I set a target srs.
> But must admit that the current behaviour was confusing. I knew how to
> interpret the input data - under the assumption that a geotransform is not
> changed when I do no transformations or interpolations:
> If I know that my data are "points" rather than "pixels" - the first
> "point" is lying half a cellsize downwards and rightwards from the upper
> left corner that GDAL reports. That's what I always assume - not relying on
> other metadata items than the GeoTransform.
> 
> So I agree that it might be worth a ticket.

Please file one.

> 
> Cheers,
> Simon
> 
> On Thu, Nov 19, 2015 at 10:29 AM, Ari Jolma <ari.jolma at gmail.com> wrote:
> > On 19.11.2015 11:26, Even Rouault wrote:
> >> GDAL version:
> >> GDAL 1.11.2, released 2015/02/10
> >> Simon,
> >> 
> >> Regarding the fact that the geotiff file looses the half-pixel shift, it
> >> seems
> >> to be due to the absence of a SRS. If I do "gdal_translate test.asc
> >> test.tif -
> >> a_srs EPSG:32630" for example, then the geotransform I get back is
> >> correct.
> >> This is a defect of the GeoTIFF writer that writes the GeoTIFF Tiepoints
> >> with
> >> the PixelIsPoint convention (since your original data is flagged as
> >> such), but
> >> fails to write GTRasterTypeGeoKey=RasterPixelIsPoint as a geotiff key
> >> when there's no SRS. Could be worth a ticket.
> > 
> > My conclusion is the same. I think the key should be set regardless of
> > the existence of a SRS.
> > 
> > Ari
> > 
> >> Some background:
> >> https://trac.osgeo.org/gdal/wiki/rfc33_gtiff_pixelispoint
> >> 
> >> A workaround :
> >> gdal_translate test.asc test.tif -mo AREA_OR_POINT=AREA
> >> 
> >> This will force the geotiff writer to use the default convention, and
> >> thus avoiding going back to the centerofpixel convention when writing
> >> the GeoTIFF
> >> tiepoints.
> >> 
> >> Even
> > 
> > _______________________________________________
> > gdal-dev mailing list
> > gdal-dev at lists.osgeo.org
> > http://lists.osgeo.org/mailman/listinfo/gdal-dev

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com