[Digikam-users] Migrating to digiKam from Windows (story) - metadata issue

[Martin Javorek]

Hi all digiKam users (and developers),

I just realized, that I must try this amazing piece of software. I had no idea, that it is already ported to windows more than 2 years. 

So, as a windows user, I tried to start there. But digiKam 1.7 was not so successful, some crashes while reading my photos (maybe I'll report them later, at first I tried to run the digiKam), database went 3 times away (SQlite) when digiKam crashed. I tried to play with MySQL, but not sucessfull with switching to this DB in digiKam on Win.    

OK, virtualbox, Ubuntu 10.10 - I want you digiKam! :-)

Installed digiKam version 1.8, playing 2 days with MySQL database (to avoid problems with crashed SQlite), but some performance problem appeared (same as with the previous issue, maybe I'll report them later - I detected some SQL queries had problems as they were about 40s per query). 

Back to SQlite, hope, digiKam will be on Linux more as fish in the water (better than on Win) and will not crash so often. Also the support - community is wider.

And then metadata. I'm tagging all my photos, sometimes filling also headline/caption. First problem - character encoding of IPTC. I'm using national characters (cp1250). I can reencode whole photoarchive with exiftool to UTF-8. OK, IPTC was not originally designed to handle UTF-8 characters, but there is a small workaround and there are several softwares, which are using this flag and can work with this. Unfortunately, digiKam cannot and which is worse, will not: http://bugs.kde.org/show_bug.cgi?id=132244 (see the end of the story). XMP is here about ten years and the switch from IPTC to XMP is not still done (and still very painfull). Maybe it will continue next ten years, nobody knows. This small flag (and small fix in source code) will help with compatibility metadata (my opinion).

Ok, let's throw away IPTC. But, how can I control digiKam not to write any IPTC data? Many software are reading primary IPTC and after that secondary XMP. If digiKam cannot handle national characters, it would be fine not to break IPTC with bad characters and have an option not to write IPTC at all. Or - some possibility to hook user script "after metadata save", I can then prepare exiftool IPTC remover for myself.

Let's go next. I can forgot IPTC if I have XMP. But now - where my photo "headline field" is? According to many reccomendations and IPTC/XMP specifications (nice collection of links http://forum.galleryserverpro.com/default.aspx?g=posts&t=1086), I'm using IPTC/XMP:headline field for brief photo description (few words) and if I need to write more, I use description/caption field. DigiKam use only caption field. Headline is editable only in detail metadata popup and no easy synchro between IPTC/XMP then. Which is not so handy, if I need in 100% cases to enter headline and in 2% to enter description/caption... 

There is a big mess around the title/headline/caption/description between various software and web galleries (commercial also). But it would be fine to have possibility to input these basic metadata more comfortable.


The only (maybe) solution for me is:

1) Forgot IPTC, have only XMP.
2) Exchange headline with description in all my photos (exiftool is a magic) to have comfortable entering headline in digiKam. When I need to exchange them back in future, I can.   
3) When I'll be changing anything in my photos using another SW than digiKam, I need to clear IPTC before (or create new IPTC copying from valid XMP), not to confuse that software with broken characters written by digiKam. This is not easy as I have 20K+ photos and I must detect, which IPTC I will be repairing (not one by one, but also not all at once).  
4) Stick with Linux version of digiKam as there are newer versions and wider community for help. 

I love digiKam functions like lighttable or facetagging or XMP sidecars (in 2.0, I'm looking forward to) and many others.

But, a little bit complicated start and workflow, aren't they?

Martin