How Aperture encodes non-ascii chars

Question

Level 1

33 points

How Aperture encodes non-ascii chars

It looks like Aperture uses a way of encode chars as UTF that seem to be bit unusual, many application (especially web apps) expects another way of encoding Unicode chars. The end result is that non-ascii chars comes out really strange.

When I use applescript to read the IPTC info I need to do something like this before using the data:

set c_title to normalize unicode p_title without decomposition

Is there some way of telling Aperture to do this automatically when exporting photos?

(sorry about the bad use of Unicode terminology but I don't know the area very well)

various, Mac OS X (10.6.6)

Posted on Jan 30, 2011 1:11 AM

Reply

Answer 1

Tom Gewecke

Level 10

118,704 points

Jan 30, 2011 5:30 AM in response to Vittra

Could you give a specific example, like a url, which someone else can view to see what you are talking about. Normally it should not make any difference on the web whether Unicode characters appear in composed or decomposed form. (The file system used by OS X only accepts decomposed for filenames).

Reply

Answer 2

Vittra Author

Level 1

33 points

Jan 30, 2011 5:43 AM in response to Tom Gewecke

Sure, here is an example using Wordpress

http://mostrom.eu/en/2011/01/30/example-of-encoding/

Reply

Answer 3

Tom Gewecke

Level 10

118,704 points

Jan 30, 2011 6:07 AM in response to Vittra

It looks like the correct UTF-8 is being read as Latin-1 encoding. The combining umlaut is UTF-8 bytes CC 88, and this is being translated to Latin-1 CC = Ì (Latin-1 88 is not a character). I don't know where or why this is happening...

Reply

Answer 4

Mar 31, 2011 1:59 PM in response to Vittra

This bug is now affecting me and is a major problem for me. Everything I export and load up to my agency's website comes out with jumbled text.

Apple needs to fix this ASAP!

Reply

Answer 5

tilman

Level 4

1,235 points

Mar 31, 2011 2:46 PM in response to Ottaviano

I've spent quite a bit of time on this topic - I'm the author of Phoshare http://code.google.com/p/phoshare/, an open-source tool to export images and metadata from iPhoto and Aperture. Storing non-ASCII characters in image metadata is a messy business. If you want to get a taste of it, have a look at the Exiftool FAQ at http://www.sno.phy.queensu.ca/~phil/exiftool/faq.html#Q10 . A quote: "Most textual information in EXIF is stored in ASCII format, ... However *it is not uncommon for applications to write UTF‑8 or other encodings where ASCII is expected*". This leads to all sorts of problems, because the reader will have to make assumptions about the encoding used by the writer.

In the case of Aperture, I have found that it writes metadata encoded in a way consistent with what most other applications and on-line services expect. Most of the encoding problems I've debugged where caused by bad input data. E.g. the characters were encoded improperly to begin with, but in a way that they still display as expected in some places (like Aperture). Can you try erasing some of the meta data in Aperture, and retype them from scratch (e.g. not copy-paste, which would paste the same stuff right back). Also make sure that you force-updated the preview images, so that the new meta data get written into the files.

Reply

Answer 6

Mar 31, 2011 11:22 PM in response to tilman

I can try and export some more images and pass them along to you. Is there a way I can contact you directly?

I use a text replacement tool to speed up captioning, but I could export several images where I try entering my captions in several different methods and see if there is a difference.

Reply

Answer 7

Apr 1, 2011 12:16 AM in response to Ottaviano

It seems to me that Aperture is assigning "Mac Roman" instead of "Latin 10 Scandinavian" I see it fine on my system but on other systems it becomes corrupted.

Reply

Answer 8

tilman

Level 4

1,235 points

Apr 3, 2011 12:01 AM in response to Ottaviano

I think the only useful encodings for metadata are Ascii or UTF-8. I don't think there is a way to make Aperture use "Latin 10 Scandinavian", and even if you could, I don't think other programs would be able to decode such strings. Maybe you can share some sample images with me, and I have a look. E-mail to tsporkert at gmail.com.

Reply

Answer 9

Nov 29, 2011 3:37 AM in response to tilman

Other professional grade image programs, like Photo Mechanic, let you set the encoding on IPTC meta data. Right now I'm working around this nightmare of a bug by exporting my images and then adding captions in Photo Mechanic, which lets you set the character encoding on IPTC meta data.

I'm lossing tons of valueable time on the field I might add and I would love for Apple to at least give some kind of acknowledgement to the issue. If I need to wait and eventually buy Aperture 4 for a fix I'm fine with it, but at least address and acknowledge this problem.

Reply

Answer 10

Tom Gewecke

Level 10

118,704 points

Nov 29, 2011 5:01 AM in response to Ottaviano

If you haven't already done so, make sure you tell them directly:

http://www.apple.com/feedback/aperture.html

Reply

Answer 11

Nov 29, 2011 5:09 AM in response to Tom Gewecke

Thanks for the suggestion Tom, I've sent feedback regarding this bug at least 3 or 4 times this year. It really makes me wish Apple could strike a little better balance between maintaining secrecy and providing information to working professionals that use their software.

I've even sent Tim Cook an email about this issue on the off chance that it get's through his mail filters and he might read it and pass it to the Aperture team.

Reply

Answer 12

Dec 9, 2011 3:40 AM in response to Vittra

Aperture 3.2.2 was released today and I was disapointed to discover this bug has still not been address. What's a guy got to do to get this address? This app is marketed to professional photographers and the tagging system is useless to anyone using any accented letters.

Reply