[gdal-dev] Re: gdal_translate: a preliminary `-gcp_file' patch

[Ivan Shmakov]

>>>>> Frank Warmerdam <warmerdam at pobox.com> writes:

 >> Since the current version of GDAL doesn't support reading GCPs from
 >> some of the data sets I use [1], I have to use `gdal_translate' to
 >> specify all the GCPs, like:

 > Ivan,

 > The --optfile was already mentioned.

	Yes, `--optfile' may suit my needs.  However, I'm having some
	considerations regarding `-gcp' and the GDAL CLI in general that
	I'll present separately.

 > I just wanted to add that there are dangers to trying to attach large
 > numbers of GCPs to a GeoTIFF image.  Some sort of problem with the
 > maximum tag size I believe.

	The output format doesn't matter to me now.  If there's an other
	format suitable for storing GCPs with `gdal_translate', I could
	easily switch to it.

 > So translating full density geolocation arrays to GCPs is likely to
 > cause a variety of problems.

	Actually, I wonder, whether a lot of GCPs would cause any
	problems with `gdalwarp'?

	BTW, `hdfdump' (which I've used in the original post's example)
	supports dumping a sparse subset of the HDF4 field, so it's not
	really a problem to make a file with sparse GCPs.

 > One approach is to subsample them.  Another is to treat them as
 > geolocation arrays in the sense described in:

 > http://trac.osgeo.org/gdal/wiki/rfc4_geolocate

	Isn't it more appropriate for HDF-EOS GCPs to be represented as
	specified there?

	Is there any way to convert ``ordinary'' GCPs into RFC 4 GCPs
	and vice versa?