Charset conversion not working with -n option

[mstorhas]

Hello Phil,

I'm using the CodedCharacterSet tag to convert IPTC content from Latin to UTF-8 for our customer, and it works well so far, e.g.

Code Select Expand

exiftool -charset Latin -iptc:Caption-Abstract="#äöü ÄÖÜ ß@€# #!§$%&/()=?#" -iptc:codedcharacterset="utf8" -o dest.jpg orig.jpg
However, when at the same time print conversion is disabled (I specify resolution units in numbers) using the -n option, IPTC content will still be Latin in the output file, example:

Code Select Expand

exiftool -charset Latin -iptc:Caption-Abstract="#äöü ÄÖÜ ß@€# #!§$%&/()=?#" -iptc:codedcharacterset="utf8" -JFIF:ResolutionUnit=1 -o dest.jpg -n orig.jpg
Instead of utf8 I also tried "ESC % G" for CodedCharacterSet, but it didn't work either. I could relatively easy get around this by adding # to my resolution unit tags and removing the -n option, but I wanted to report this behaviour nevertheless.
I'm using command line on windows xp and tried version 9.02.

Greetings, Martin

[Phil Harvey]

Hi Martin,

The problem is that you must specify the binary escape sequence for CodedCharacterSet when -n is used (the same for reading or writing):

Code Select Expand

> exiftool a.jpg -codedcharacterset=utf8
    1 image files updated

> exiftool a.jpg -codedcharacterset
Coded Character Set             : UTF8

> exiftool a.jpg -codedcharacterset -n
Coded Character Set             : .%G

> exiftool a.jpg -codedcharacterset -b | hexdump -
    0000: 1b 25 47                                        [.%G]

This may seem odd, but it allows full control over the escape sequence if necessary.

- Phil