Hello Jukka
your assumption how the the OFFSITE tag is applied for the WCS output is
also what I would have expected it to do. I probably did not notice this
strange behaviour since my set ups of WCS only used non-overlapping
tiles (for WMS I use overlapping tiles and the OFFSITE works fine).
Just an idea (I never tried it): maybe you could try to use a VRT image
instead of the tileindex. At least since Gdal 1.7 there is the tool
"gdalbuildvrt" that has the options "-srcnodata" and "-vrtnodata" which
might do what you want.
All tests I made so far with VRT images worked quite fine, they were
exactly treated as a single image. I don't know if the speed might be a
bit slower, but WCS is typically slow anyway, so it should not cause
problems.
Greetings
Armin
On 19/04/2010 16:10, Rahkonen Jukka wrote:
Frank Warmerdam wrote:
Jukka,
Generally WCS generated geotiffs will not have a way to
represent nodata
in a way that a client application is likely to recognise
them as nodata.
By default pixels which are assigned no value from the source
dataset are
set to zero.
I am not sure if I could tell clearly what was my problem. I will try
again. I have layers like this:
LAYER
NAME "M138"
METADATA
"wcs_label" "M138"
"OWS_EXTENT" "86000 6636000 650000 7518000"
"WCS_RESOLUTION" "0.5 0.5"
"WCS_SIZE" "50000 50000"
"WCS_BANDCOUNT" "1"
END
STATUS ON
TILEINDEX "d:/dataM138"
TYPE RASTER
UNITS METERS
SIZEUNITS PIXELS
OFFSITE 0 0 0
TOLERANCE 3
TOLERANCEUNITS PIXELS
PROJECTION
"init=epsg:3067"
END
DUMP TRUE
END
Tiles combined by M138_WMS.shp overlap because they are reprojected.
When I use this layer through WMS, having OFFSITE 0 0 0 makes the black
nodata triagles transparent and warped images make a seamless mosaic.
Somehow I was awaiting that WCS behaves in a similar way, but it does
not. Black nodata triangles are burned into the output image and the
mosaic is useless. I can get my job done by hand editing WMS calls so
that the BBOX and WIDTH and HEIGHT give me the desired pixel size, but
it would be a bit more simple through WCS GetCoverage and ResX, ResY.
-Jukka-
_______________________________________________
mapserver-users mailing list
mapserver-users@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/mapserver-users
_______________________________________________
mapserver-users mailing list
mapserver-users@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/mapserver-users