Opened 18 years ago

Last modified 17 years ago

#1225 closed defect

gdal_merge.py creates as output non-standard GeoTIFF images — at Version 2

Reported by: gregcoats@… Owned by: warmerdam
Priority: normal Milestone: 1.4.2
Component: GDAL_Raster Version: unspecified
Severity: normal Keywords:
Cc: gregcoats@…

Description (last modified by warmerdam)

When using gdal_merge.py to mosaic together a series of smaller GeoTIFF images, the gdal_merge.py created GeoTIFF output is non-standard. For one example, feeding the gdal_merge.py created GeoTIFF as an input to geotifcp to create as output a tiled GeoTIFF produces an immediate Segmentation fault. For another example, the file size of the gdal_merge.py created GeoTIFF is much larger than necessary: in one case the gdal_merge.py created GeoTIFF was 84,656,520 bytes but simply copying the GeoTIFF with geotifcp without specifying any option yields a proper GeoTIFF of only 60,933,970 bytes. In another case, specifying gdal_merge.py -co "INTERLEAVE=PIXEL" resulted in gdal_merge.py creating a non-standard GeoTIFF of 127.8 MB, that Photoshop was able to open and save without compression in only 58.1 MB. I also found that specifying gdal_merge.py -co "INTERLEAVE=PIXEL" -co "TILED=YES" -co "BLOCKXSIZE=256" -co "BLOCKYSIZE=256" resulted in a visibly corrupt image, with errant red, green, and blue pixels scattered in various pats of the image. I access GDAL through William Kyngesburye's GIS Libs 3.1.1, under Mac OS X 10.4. For now, the work around is to use gdal_merge.py without specifying any options, and then immediate feeding that as an input to gdal_translate and specify the desired -co options.

Greg Coats

Change History (1)

comment:2 by warmerdam, 17 years ago

Component: PythonBindingsGDAL_Raster
Description: modified (diff)
Milestone: 1.4.1
Priority: highnormal
Severity: majornormal

Greg,

Per bug #1470, I recently fixed some issues with file bloat and the geotiff writer.

Please retest with GDAL trunk (or a nightly snapshot) and see if the problem still occurs. If so, I will have Mateusz try and reproduce the problem and we can go from there.

I am reclassifying this as GDAL_Raster since I don't see any evidence the problem relates to the python bindings.

Note: See TracTickets for help on using tickets.