Extracting IPTC from various pics fails

[holger]

I am using Imagemagick 6.5.7 in the moment , but I was able to test and verify the following on IM 6.4.0, IM 6.4.2, IM 6.4.6 and IM 6.5.3

ImageMagick is not able to extract IPTC information from pictures. According to my tests this failure is related to the XMP format used in different Adobe apps. While the IPTC information from Photoshop CS3 with Adobe XMP Core 4.1-c036 can be extracted it is not possible to read the IPTC from Bridge CS3 and Bridge CS4 (Adobe XMP Core 4.2-c020) or Photoshop CS4 (Adobe XMP Core 4.2.2-c063).

Pics for testing this issue can be provided.

[magick]

Post a URL to a few sample images so we can reproduce the problem.

[holger]

ftp://ftp.prhamburg.com/testfiles/IM_IPTC_XMP.zip

Filenames should be self-explanatory:
_MG_4082BridgeCS3.tif
_MG_4082BridgeCS4.tif
_MG_4082PhotoshopCS3.tif
_MG_4082PhotoshopCS4.tif


Only _MG_4082PhotoshopCS3.tif works correctly with convert filename 8BIMTEXT:-

[magick]

Can you define "works correctly"? We can extract the 8BIM profile from each of the TIFF images you provided with

• convert image.tif 8BIMTEXT:-

We're using ImageMagick 6.5.8-0 and libtiff 3.9.1.

[tko]

"works correctly" means the IPTC-fields are missing:

## OK-image:
..$ convert _MG_4082PhotoshopCS3.tif 8BIMTEXT:-
8BIM#1028="IPTC"
2#0="�"
2#120#Caption="Lilly findet Ostergeschenke Photoshop"
2#122#Caption Writer="Marc Photoshop"
2#105#Headline="Wer suchet, der findet. Photoshop"
2#40#Special Instructions="Dieses Bild bitte nicht drucken Photoshop"
...
2#25#Keyword="Ostern"
2#25#Keyword="Garten Photoshop"
2#92="zu Hause im Garten Bridge"
8BIM#1061="‰o=ï°Ì¿ ˜J3"
....

## bad image
..$ convert _MG_4082BridgeCS3.tif 8BIMTEXT:-
8BIM#1061="ÁÄÎNr Ìžˆ=×z:×"
...
-> output directly starts with 8BIM#1061

I could reproduce this with versions ImageMagick 6.5.8-2 2009-12-01 Q16 and LIBTIFF, Version 3.9.2, so a little later&greater than your suggestion.
Opening the images as text show all 3 "bad" ones have "Adobe XMP Core 4.2" or 4.2.2. The OK.version has "Adobe XMP Core 4.1". So this is suspect somehow...

[holger]

tko's findings are identical with mine!

[tko]

mh, I think I was on the wrong track:
with other examples I found OK-images with XMP 4.2. The info is obviously not taken from XMP(notaton) but from the "real" IPTC-Header. For "bad images" I found this was written to the end of the file, so you can´t really call that a header any more... see also the examples from holger.