Even Rouault wrote > Le jeudi 08 août 2013 19:44:04, Hermann Peifer a écrit : >> On 2013-08-08 9:55, ludwig.hilger wrote: >> > Hello everybody, >> > >> > I am also a newbie and seem to have a similar problem, but cannot get >> it >> > solved. I am trying to reproject the following file: >> > http://www.altmuehlnet.de/~hilger/Test_Smooth_0208_4258.tif >> > I am using the following line: >> > gdalwarp -et 0 -r bilinear -s_srs EPSG:4258 -t_srs EPSG:25832 >> > Test_Smooth_0208_4258.tif Test_Smooth_0208_4258_25832.tif >> >> I downloaded your test file and noted that it has 10000x10000 pixel and >> a real world extent of around 7x7 metres. So 1 pixel in your GeoTIFF >> represents less than 1x1 mm. I can only blindly assume that some >> rounding effects related to the small pixel size lead to the problem you >> observed. I noted the same problem in my testing (I am using the MacPort >> of GDAL 1.10, on my MacBook). >> >> I re-defined the extent of your file to 1x1 degree and was able to warp >> the resulting file without any problem: >> >> $ gdal_translate Test_Smooth_0208_4258.tif out_1x1.tif -a_ullr 10 47 11 >> 46 >> >> $ gdalwarp out_1x1.tif out_25832.tif -t_srs epsg:25832 -et 0 -rb >> >> Maybe someone from the GDAL devs has an explanation for this behaviour. > > Herman, > > you're right. The issue was due to the input pixel size being very small. > I've > fixed that in ticket http://trac.osgeo.org/gdal/ticket/5190
I should have seen in the DEBUG output that before your fix, always the same source pixel (Src=0,0,1x1 - which happened to have the value 0 in all bands) was taken and warped into the various destination windows. Now it works as expected. Thanks, Hermann -- View this message in context: http://osgeo-org.1560.x6.nabble.com/gdalwarp-produces-all-black-output-tp5019465p5071685.html Sent from the GDAL - Dev mailing list archive at Nabble.com. _______________________________________________ gdal-dev mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/gdal-dev
