[gdal-dev] Working with NTF files (2)

[Nikos Alexandris]

On Wednesday 31 of July 2013 11:55:27 Nikos Alexandris wrote:
> Yet another NITF related "problem". An ntf file contains (no SUBDATASETS)
> 
> gdalinfo -nogcp -nomd -mm M1BS.ntf
> 
> ,--%<---
> Band 1 Block=1024x1024 Type=UInt16, ColorInterp=Blue
>     Computed Min/Max=183.000,2047.000
>   Overviews: 3438x3448, 1719x1724, 859x862, 429x431, 214x215
>   Overviews: arbitrary
> Band 2 Block=1024x1024 Type=UInt16, ColorInterp=Green
>     Computed Min/Max=252.000,2047.000
>   Overviews: 3438x3448, 1719x1724, 859x862, 429x431, 214x215
>   Overviews: arbitrary
> Band 3 Block=1024x1024 Type=UInt16, ColorInterp=Red
>     Computed Min/Max=86.000,2047.000
>   Overviews: 3438x3448, 1719x1724, 859x862, 429x431, 214x215
>   Overviews: arbitrary
> Band 4 Block=1024x1024 Type=UInt16, ColorInterp=Alpha
>     Computed Min/Max=50.000,2047.000
>   Overviews: 3438x3448, 1719x1724, 859x862, 429x431, 214x215
>   Overviews: arbitrary
> `--->%--
> 
> 
> Attempting to warp the data with the -rpc switch and convert to GeoTIFF,
> begins with
> 
> "Using band 4 of source image as alpha."
> 
> 
> This results in an "empty" NIR band  in the GeoTIFF file.
> 
> ,--%<---
> Band 1 Block=6728x1 Type=UInt16, ColorInterp=Gray
>     Computed Min/Max=0.000,2047.000
> Band 2 Block=6728x1 Type=UInt16, ColorInterp=Undefined
>     Computed Min/Max=0.000,2047.000
> Band 3 Block=6728x1 Type=UInt16, ColorInterp=Undefined
>     Computed Min/Max=0.000,2047.000
> Band 4 Block=6728x1 Type=UInt16, ColorInterp=Undefined
>     Computed Min/Max=0.000,0.000
> `--->%--
> 
> 
> No success using '-co "ALPHA=NO"'.  Same "problem" when using gdal_translate
> on the whole ntf file. However, instructing
> 
> gdal_translate -b 4 M1BS.ntf b4.tif
> 
> 
> gives
> 
> gdalinfo -nogcp -nomd -mm b4.tif
> 
> ,--%<---
> Band 1 Block=6876x1 Type=UInt16, ColorInterp=Gray
>     Computed Min/Max=50.000,2047.000
> `--->%--
> 
> 
> But, this is not "fine". How is this to be done correctly, RPC-based warping
> and conversion in one go?

Related thread/solution:

http://lists.osgeo.org/pipermail/gdal-dev/2012-December/034984.html

Translating to a VRT, band by band -- is it the only solution?

Nikos