[gdal-dev] Blocks of data missing when using gdalwarp

Chaitanya kumar CH chaitanya.ch at gmail.com
Wed Mar 4 12:59:06 EST 2009


Asger,

Can you provide the source tif files to reproduce the problem?

Regards,
-- 
Chaitanya kumar CH.

On Wed, Mar 4, 2009 at 11:17 PM, Asger Sigurd Skovbo Petersen
<ASP at blominfo.dk> wrote:
> Hi list,
>
> I have a setup for transforming a large quantity of raster tiles from utm into spherical mercator. The setup is like:
>
> -------------------
>
> // This first step has some rather unimportant explanations foreach SourceFile.tif {
>  gdal_translate -of vrt SourceFile.tif SourceFile.vrt }
>
> gdalbuildvrt SourceMosiac.vrt SourceFile*.vrt
>
> foreach outputtile
> {
>  gdalwarp -te tilebounds -of gtiff -co tiled=yes -co interleave=band
>        -t_srs "+proj=merc +a=6378137 +b=6378137 +lat_ts=0.0 +lon_0=0.0
>        +x_0=0.0 +y_0=0 +k=1.0 +units=m +nadgrids=@null +no_defs" -r cubic -wm
>        100 SourceMosaic.vrt outputtile.tif
> }
>
> -----------------------
>
> This _almost_ works as a dream. The only problem is that the warped tif tiles have some very systematic blocks of black pixels which suspiciously look like warp errors. They are definitely not in the input data. The blocks are strictly rectangular and they are found in patterns that are very much alike from outputtile to outputtile.
> Please see attached jpg file for an example.
>
> Has anyone seen this before and are there any ways to work around this weird effect?
>
> Best regards
> Asger
>
>
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/gdal-dev
>


More information about the gdal-dev mailing list