[gdal-dev] alpha mask to data value

[William Kyngesburye]

On Apr 2, 2014, at 4:33 PM, Even Rouault <even.rouault at mines-paris.org> wrote:

> Le mardi 01 avril 2014 22:14:11, William Kyngesburye a écrit :
>> I've wrestled with various nodata issues in the past, now it's hitting me
>> again...
>> 
>> I'm using Photoshop to delete collars on scanned maps, creating an alpha
>> mask.  GDAL has no problem with this.  What I want to do is merge maps
>> together (after rectification), then set any remaining nodata areas to
>> white, RGB 255,255,255.  Just dropping the extra alpha band doesn't work
>> because nodata is set to 0,0,0, which is black.  The a_nodata option in
>> gdal_translate just defines what existing value in the data is nodata.
>> 
>> I tried using gdalwarp with the -dstnodata option which should set nodata
>> values in the output to a specific value, but it carries along the alpha
>> band and ignores dstnodata.
>> 
>>  gdalwarp -dstnodata "255 255 255" in.tif out.tif
>>  Processing input file in.tif.
>>  Using band 4 of source image as alpha.
>>  Using band 4 of destination image as alpha.
> 
> You can perhaps try adding  -wo INIT_DEST="255,255,255,0"


That appears to work, though not immediately.  The raster has 255 in the nodata areas, but it's not until I strip out the alpha band with gdal_translate that it shows.

Now I see an odd artifact, or maybe bug somewhere - with nodata values in an RGB image, if any one or two of the bands for a pixel are 255 while the other bands are not, the pixel as a whole is treated as nodata when displaying in QGIS.  Though identifying such a pixel will show nodata for just the band(s) that are 255.  Photoshop has no concept of nodata values, so it displays correctly (and with white for fully nodata pixels).

I think this explains another issue I saw at one point when I had ovelap when merging 2 rasters: the pixels of the top raster covered the pixels of the lower at the band level, which caused the band of the lower to show through when the upper band was 255, and resulted in a wild color mix. ie:

result 127 127 127 (grey) but should have been 255 127 127
        ^
upper  255 127 127 (pink)
        ^
lower  127 127 127 (grey)

I think I'll save the nodata flattening to white until the end, after merging.

-----
William Kyngesburye <kyngchaos*at*kyngchaos*dot*com>
http://www.kyngchaos.com/

Earth: "Mostly harmless"

- revised entry in the HitchHiker's Guide to the Galaxy