JPEG 12-bit compression for NITF Files about gdal

[rouault]

@chuck1987 What are the minimum and maximum values reported by the -mm option:

gdalinfo.exe -mm .\11JAN15WV010400015JAN11050953-P1BS-500827961040_01_P005_________AAE_0AAAAEAQABC9.NTF

[chuck1987]

@rouault

Band 1 Block=256x256 Type=UInt16, ColorInterp=Gray
Computed Min/Max=0.000,1268.000
Overviews: 8704x12800, 4352x6400, 2176x3200, 1088x1600, 544x800, 272x400, 136x200
Metadata:
NITF_ISUBCAT=651
Image Structure Metadata:
NBITS=11

[rouault]

Weird, the min/max reported is fine. I can't make sense why you get the "One or more pixels clipped to fit 12bit domain for jpeg output." warning. I'd need access to the input file to investigate further

[chuck1987]

@rouault That was the first thing I found strange, I even went as far as converting the NITF to a Byte data-type and then trying this, but I get the same warning and error. I assumed by looking at the source, that even if the maximum was greater then 12bit, then we would just clamp the value and it should still work? I was able to get the JPEG 12-bit conversion to work for more basic Single Band Byte NITFs. Any NITF I have with 11-bit JPEG2000 compression however, will not work. The file is to large to add through github, what would be the most convenient way for you to get the file?

[rouault]

The file is to large to add through github, what would be the most convenient way for you to get the file?

dropbox or equivalent file sharing services

[chuck1987]

The file can be downloaded from here.

https://hawk.compusult.net/gdal_jpeg12_issue.zip