[gdal-dev] gdalwarp fills nulls?

[Even Rouault]

Selon Greg Coats <gregcoats at mac.com>:

-wm is an option specific to gdalwarp. Here's detailed explanation :
http://trac.osgeo.org/gdal/wiki/UserDocs/GdalWarp#WarpandCacheMemory:TechnicalDetails

> The GDAL default value for GDAL_CACHEMAX is 40 MB. I routinely set up
> my gdal_translate scripts to run with a larger value for GDAL_CACHEMAX.
> 	export GDAL_CACHEMAX_value=50
> 	gdal_translate --config GDAL_CACHEMAX ${GDAL_CACHEMAX_value}
> In my benchmark tests, I have observed that this small increase in
> GDAL_CACHEMAX has dramatically reduced processing time by a factor of
> more than 60, from 10473 to 166 seconds.
>
> I do not see any information about a -wm option on these web pages
> 	http://www.gdal.org/gdal_utilities.html
> 	http://www.gdal.org/gdal_translate.html
> I would appreciate information about the -wm option. What is the
> default value? Are the units MB of memory?
> Greg
>
> On Sep 11, 2009, at 11:42 AM, William Kyngesburye wrote:
>
> > Hmm, yes that one works for me.  I simplified my example for the
> > email, maybe there's a memory problem, or raster size problem?
> >
> > I originally tried to resample a GRASS raster DEM directly.  It's
> > really 1000m cell size, 12809x18192 float cells, and want to
> > resample to 21000m and 7000m.  That wasn't working so I exported to
> > Geotiff (Int16 cells) with r.out.gdal.  The exported image has the
> > nulls as -32768.  That doesn't work either.
> >
> > ... ah! memory!  I just tried with -wm 250 and I get the nulls.  And
> > the nasty artifacts.  I think I want the nulls filled now ;)  I saw
> > a bug about this on the tracker but it was fixed.  Maybe this is a
> > different problem?
> >
> > On Sep 10, 2009, at 5:56 PM, Even Rouault wrote:
> >
> >> Selon William Kyngesburye <woklist at kyngchaos.com>:
> >>
> >> I don't manage to reproduce your issue neither with trunk nor with
> >> 1.6.2 with a
> >> indem.tif generated by the following python script :
> >>
> >> import gdal
> >> import osr
> >>
> >> ds = gdal.GetDriverByName('GTiff').Create('indem.tif', 500, 500, 1,
> >> gdal.GDT_Int16)
> >> ds.SetGeoTransform([500000, 10, 0, 45000000, 0, -10])
> >> sr = osr.SpatialReference()
> >> sr.ImportFromEPSG(32631)
> >> ds.SetProjection(sr.ExportToWkt())
> >> ds.GetRasterBand(1).Fill(-32768)
> >> ds.GetRasterBand(1).SetNoDataValue(-32768)
> >> ones = ''
> >> for i in range(50*50):
> >>   ones = ones + '\001\000'
> >> ds.GetRasterBand(1).WriteRaster(225,225,50,50, ones, 25, 25)
> >> ds = None
> >>
> >>
> >> and
> >>
> >> gdalwarp -tr 100 100 -ot Int16 -r lanczos -dstnodata -32768
> >> indem.tif outdem.tif
> >>
> >> You need to specify -dstnodata -32768. The result I get is as
> >> expected a square
> >> of ones at the middle of nodata values at -32768 (except that due
> >> to a strange
> >> behaviour of lanczos you get some artifacts at the edge of the
> >> square).
> >>
> >>
> >>
> >>> Maybe I'm missing something?  I'm trying to resample a DEM, with
> >>> null/
> >>> nodata values in oceans, using gdalwarp.  No matter what options
> >>> I've
> >>> used so far, all the resampled nodata areas get filled with 0.
> >>>
> >>> GDAL 1.6.2
> >>>
> >>> So far I've tried (src res 10, nodata -32768 defined in src):
> >>>
> >>> gdalwarp -tr 100 100 -ot Int16 -r lanczos indem.tif outdem.tif
> >>> gdalwarp -tr 100 100 -ot Int16 -r lanczos -dstnodata -32768
> >>> indem.tif
> >>> outdem.tif
> >>> gdalwarp -tr 100 100 -ot Int16 -r lanczos -dstalpha indem.tif
> >>> outdem.tif
> >>>
> >>> The last time I used gdalwarp, with nulls, was back at 1.5.x, and it
> >>> worked then, though I was projecting data (-t_srs), instead of only
> >>> resampling.  (I'd have to dig up an old 1.5 to see if it still
> >>> works)
> >>>
> >>> -----
> >>> William Kyngesburye <kyngchaos*at*kyngchaos*dot*com>
> >>> http://www.kyngchaos.com/
> >>>
> >>> "I ache, therefore I am.  Or in my case - I am, therefore I ache."
> >>>
> >>> - Marvin
> >>>
> >>>
> >>> _______________________________________________
> >>> gdal-dev mailing list
> >>> gdal-dev at lists.osgeo.org
> >>> http://lists.osgeo.org/mailman/listinfo/gdal-dev
> >>>
> >>
> >>
> >
> > -----
> > William Kyngesburye <kyngchaos*at*kyngchaos*dot*com>
> > http://www.kyngchaos.com/
> >
> > "History is an illusion caused by the passage of time, and time is
> > an illusion caused by the passage of history."
> >
> > - Hitchhiker's Guide to the Galaxy
> >
> >
> > _______________________________________________
> > gdal-dev mailing list
> > gdal-dev at lists.osgeo.org
> > http://lists.osgeo.org/mailman/listinfo/gdal-dev
>
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/gdal-dev
>