On 28-06-2010 23:21, Michael Sumner wrote:
Am I correct in interpreting this post to mean that the pixel-is-area and pixel-is-point tiffs>are
read the exact same way by GDAL, with the only difference being in the value of the>AREA_OR_POINT
metadata key? Should viewers then be looking for>the "AREA_OR_POINT" key in addition to
the georeferencing information, and shift the>display to center the top-left pixel on (0,0) if
it's set to Point?
That's my understanding, it's just a metadata tag there for use if you
need it. I would say that viewers should only modify their behaviour
from Area if they have the distinction for centre point vs. cell based
views.
I've seen it used to set centre-based TIN-like renderings, vs.
cell-based image renderings - but there's nothing in the GDAL raster
storage model that prevents you taking on either interpretation AFAIK,
though the default is cell-based.
(There can be a problem interpreting more flexible array formats like
NetCDF, that can store irregularly spaced coordinate vectors for, say,
the centres of cells on the X axis - and there's rarely any metadata
(in the file) to say explicitly that it's centre- not corner-based. )
Cheers, Mike.
netCDF grids created by GMT have always (~20 years) made that
distinction and store that information.
Joaquim
_______________________________________________
gdal-dev mailing list
gdal-dev@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/gdal-dev
