Confusion over non-Exif orientation data
Posted: 2013-07-23T18:28:26-07:00
Hi,
I have a JPEG that has orientation data in it, but it's not in the EXIF data -- it's stored, AFAICT, in an XMP-tiff metadata field. When I query the file for its orientation, using "identify -format '%[EXIF:Orientation]'", it reports nothing. However, when I convert the image using "-auto-orient", it does rotate the image. The documentation for -auto-orient says that it looks at EXIF data:
http://www.imagemagick.org/script/comma ... uto-orient
but it does seem to find and act upon the orientation stored in the file, even though it's not in the EXIF data. That does seem like the right thing to do, but it's not consistent with the documentation.
In this thread on the users forum ( viewtopic.php?f=1&t=23777&p=101043#p100897 ) we discovered that using the format string "%[Orientation]" instead of "%[EXIF:Orientation]" will extract the orientation from this file, but will this orientation always be consistent with the orientation used to control the behaviour of -auto-orient?
The image in question is here: http://wootangent.net/~lsd/blah/0000221676_0.jpg
Thanks!
I have a JPEG that has orientation data in it, but it's not in the EXIF data -- it's stored, AFAICT, in an XMP-tiff metadata field. When I query the file for its orientation, using "identify -format '%[EXIF:Orientation]'", it reports nothing. However, when I convert the image using "-auto-orient", it does rotate the image. The documentation for -auto-orient says that it looks at EXIF data:
http://www.imagemagick.org/script/comma ... uto-orient
but it does seem to find and act upon the orientation stored in the file, even though it's not in the EXIF data. That does seem like the right thing to do, but it's not consistent with the documentation.
In this thread on the users forum ( viewtopic.php?f=1&t=23777&p=101043#p100897 ) we discovered that using the format string "%[Orientation]" instead of "%[EXIF:Orientation]" will extract the orientation from this file, but will this orientation always be consistent with the orientation used to control the behaviour of -auto-orient?
The image in question is here: http://wootangent.net/~lsd/blah/0000221676_0.jpg
Thanks!