[GRASSLIST:7162] grass rasters on mapserver

Hi all.
We're trying to add grass raster layers to a mapserver (UNM) application, but
we encounter problems. Our .map definition is:
LAYER
   NAME "landsat"
  TYPE RASTER
   STATUS DEFAULT
   DATA "/home/Documenti/datigrass/Toscana/PERMANENT/cellhd/sat"
END
but the raster (a Landsat map) is not displayed correctly. I assume this is
related to the mentioned 8/24 bit limitation mentioned in
http://grass.itc.it/start.html
However, I find contradictory notes about this:
- in the web page I find " GRASS raster map directly read from location must
be in 8bit", whereas in http://grass.itc.it/spearfish/README there is "Hint:
Mapserver currently only displays 8bit/24bit Integer images."
Is there a way to display directly grass rasters in full color, or are we
limited to 8bit (in this case, tiff may be a better choice, but we would
prefer to avoid data duplication).
And where exactly is the limitation? If it is in gdal, how can qgis display
the same rasters correctly?

All the best.
pc
--
Paolo Cavallini
cavallini@faunalia.it www.faunalia.it www.faunalia.com
Piazza Garibaldi 5 - 56025 Pontedera (PI), Italy Tel: (+39)348-3801953

On 6/15/05, Paolo Cavallini <cavallini@faunalia.it> wrote:

Hi all.
We're trying to add grass raster layers to a mapserver (UNM) application, but
we encounter problems. Our .map definition is:
LAYER
   NAME "landsat"
  TYPE RASTER
   STATUS DEFAULT
   DATA "/home/Documenti/datigrass/Toscana/PERMANENT/cellhd/sat"
END
but the raster (a Landsat map) is not displayed correctly. I assume this is
related to the mentioned 8/24 bit limitation mentioned in
http://grass.itc.it/start.html
However, I find contradictory notes about this:
- in the web page I find " GRASS raster map directly read from location must
be in 8bit", whereas in http://grass.itc.it/spearfish/README there is "Hint:
Mapserver currently only displays 8bit/24bit Integer images."
Is there a way to display directly grass rasters in full color, or are we
limited to 8bit (in this case, tiff may be a better choice, but we would
prefer to avoid data duplication).
And where exactly is the limitation? If it is in gdal, how can qgis display
the same rasters correctly?

GRASS raster driver in GDAL represents the color table as metadata and
QGIS knows about that and it can use it while Mapserver cannot.
I think that the right and 'quick' solution is to ask Mapserver developers to
add support for color tables passed in metadata. Frank plans some more advanced
color table model but it can take years(?). The format of of metadata
was discussed
with Frank.

Radim

All the best.
pc
--
Paolo Cavallini
cavallini@faunalia.it www.faunalia.it www.faunalia.com
Piazza Garibaldi 5 - 56025 Pontedera (PI), Italy Tel: (+39)348-3801953

Now I got it.
Thanks Radim.
All the best.
pc

At 14:24, mercoledì 15 giugno 2005, Radim Blazek has probably written:

On 6/15/05, Paolo Cavallini <cavallini@faunalia.it> wrote:
> Hi all.
> We're trying to add grass raster layers to a mapserver (UNM) application,
> but we encounter problems. Our .map definition is:
> LAYER
> NAME "landsat"
> TYPE RASTER
> STATUS DEFAULT
> DATA "/home/Documenti/datigrass/Toscana/PERMANENT/cellhd/sat"
> END
> but the raster (a Landsat map) is not displayed correctly. I assume this
> is related to the mentioned 8/24 bit limitation mentioned in
> http://grass.itc.it/start.html
> However, I find contradictory notes about this:
> - in the web page I find " GRASS raster map directly read from location
> must be in 8bit", whereas in http://grass.itc.it/spearfish/README there
> is "Hint: Mapserver currently only displays 8bit/24bit Integer images."
> Is there a way to display directly grass rasters in full color, or are we
> limited to 8bit (in this case, tiff may be a better choice, but we would
> prefer to avoid data duplication).
> And where exactly is the limitation? If it is in gdal, how can qgis
> display the same rasters correctly?

GRASS raster driver in GDAL represents the color table as metadata and
QGIS knows about that and it can use it while Mapserver cannot.
I think that the right and 'quick' solution is to ask Mapserver developers
to add support for color tables passed in metadata. Frank plans some more
advanced color table model but it can take years(?). The format of of
metadata was discussed
with Frank.

Radim

--
Paolo Cavallini
cavallini@faunalia.it www.faunalia.it www.faunalia.com
Piazza Garibaldi 5 - 56025 Pontedera (PI), Italy Tel: (+39)348-3801953