I'm dealing with jpeg2000 images that have 12 bit components and have been going through the jp2kak Kakadu driver. Earlier versions of this code truncated my 12 bit values to 8 bits, but the latest code returns all 12 bits to me, but I get my 12 bits in the lower 12 bits of the 16 bit data value. For this to be interpreted sensibly as an image, these 12 bits should either be shifted into the top 12 bits of the 16 bit data value or there should be a way to query the raster band and get its precision, independent of the type of the data value. I looked briefly at the Jasper implementation and I think it would also return the 12 bits in the low 12 bits of a 16 bit data value.
I have modified a version of the kakadu driver code to shift the bits appropriately, but am not sure what the "correct" solution to this problem really should be. Any thoughts from folks with more gdal experience? -- View this message in context: http://n2.nabble.com/jpeg-2000-component-precision-tp3260339p3260339.html Sent from the GDAL - Dev mailing list archive at Nabble.com. _______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/gdal-dev
