[Gdal-dev] large file support, gdal_merge with 8bit images (was: GDAL data model...)

[Hannu Koivisto]

Frank Warmerdam <warmerdam at pobox.com> writes:

> I would suggest using the gdal_merge.py script.  This will build
> a mosaic from a set of georeferenced input files.  You would need
> to have .wld (ESRI world files) for each of your PNG chunks for it
> to place them properly in the result.

I couldn't convince myself that gdal_merge would do the right thing
with my images so I went with the netpbm method.  I took a look at
gdal_merge code and it seemed to me that if the images are 8bit
images with colormaps, gdal does not make the result 24bit if
neccessary (not to speak about quantization which was required in
my case since I do not want a 24bit result), instead it blindly
uses the colormap of the first image.

> file.  You will likely need to assemble the files into an uncompressed
> file (perhaps of some format like Erdas Imagine - driver=HFA) and
> then convert that to a compressed GeoTIFF file using gdal_translate
> as a separate step.

I have now assembled the files into an uncompressed PNG.
Unfortunately, if I try to convert that to a compressed GeoTIFF
file using gdal_translate, it cannot open the input file because it
is larger than 4GB, i.e. it does not seem use large file support.