[gdal-dev] mask bands howto

[Vincent Schut]

Thanks for the info, both.

Major lesson: after ds creation, first close it, than open it GA_Update, 
then CreateMaskBand().

Another question: it seems (geotiff, external) masks are of dtype byte? 
So they are not *bit*masks and will use 8 times more space than strictly 
required (when no compression is used)? I thought the tiff spec allowed 
for 1-bit datatypes (mainly used for fax data probably)...
And, apart from the wasted space (I use compression anyway), I'd like to 
know the mask value conventions. I mean, a pixel in a mask could have 
values from 0-255, not? Which one is True (masked), which one is 
False(unmasked)? 'Cause if I give it a bool numpy array to write to the 
mask (using gdal_array), it complains about incompatible datatypes...

Cheers,
Vincent.


Even Rouault wrote:
> You can also take inspiration from the python autotest suite : 
> http://trac.osgeo.org/gdal/browser/trunk/autotest/gcore/mask.py
> 
> Le Wednesday 24 September 2008 16:37:27 Vincent Schut, vous avez écrit :
>> Hi,
>>
>> is there any up-to-date documentation/tutorial available on the recently
>> added mask bands feature? I found
>> http://trac.osgeo.org/gdal/wiki/rfc15_nodatabitmask, but I was hoping
>> more on something like an api tutorial. The doxygen generated stuff all
>> is empty for mask related functions...
>>
>> Specifically, I'd like to know if it is possible to (using swig python)
>> create a DS (gtiff), and then write boolean data into
>> ds.GetRasterBand(1).GetMaskBand()? Or should I call CreateMaskBand
>> first? This is not entirely clear to me...
>>
>> Vincent.
>> _______________________________________________
>> gdal-dev mailing list
>> gdal-dev at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/gdal-dev
> 
> 
>