[Digikam-users] How to remove "left behind" tag tree

Jean-François Rabasse jean-francois.rabasse at wanadoo.fr
Thu Oct 25 17:05:06 BST 2012


On Thu, 25 Oct 2012, Elle Stone wrote:

> The slowness of writing to all the image files is one reason I've been
> investigating using XMP sidecar files, as writing to a text file is
> much faster than writing to an image file.  Unfortunately, apparently
> digiKam doesn't reread the XMP files, so the XMP sidecar files can't
> be kept in synch with the database.

FYI :
Slowness problem comes from the metadata handling library, libexiv2,
used by Digikam, and DK isn't to be blamed. I have personal experience
with this library because I use it for some personal programs.
The API provides two major calls, readmetadata(), then writemetadata()
after edition. But what seems to me a design weakness is that the 
readmetadata is actually a read-and-decode, instead of a read and wait
for decoding upon first need.

So, if a program wants to edit a XMP property, e.g. the Image title,
the read will also trigger a reading of the whole Exif section and full 
decoding, including Makernotes parsing, and decoding all the stuff.
A really useless and high CPU consumming task !

But Digikam is only client of that library. Seems to be another good
reason to try to feature all things that could limit the needs for writing
to files, maintenance mode or such.


> I agree that changing your tag organization *ideally* should be done
> rarely or never. However, in reality, figuring out how to organize
> tags is an iterative learning process. No-one knows when they start
> tagging images how they really want the tags to be structured. So tag
> reorganization is essential for beginners and also for peope who've
> been tagging for a while.

And also for people that happen to import images sets that have been
tagged with another application. Digikam reads the hierarchicalSuject
fields when scanning for new images, and creates on the fly the tags
structure. But in 95% of cases, the structure set by another individual
isn't your personal structure, and you often need to reorganize, to fit
the original tags tree into yours.


> The last time I reorganized my tag tree was because digiKam doesn't
> make it very easy to search on "not any tag in this entire tag tree".
> So I broke the tag tree - which kept track of location by country,
> stateprovince, city and sublocation - into separate trees, one for
> country, one for stateprovince, etc.

A dirty trick to process missing tags :
Mark all your images with dummy tags, /Nocountry, /Nocity, etc.
Then select all images with the /Location/Country tag set, and drop the
Nocountry tag on these images. Etc.
At end, select all your images with /Nocountry, it's the « remaining »
not yet localized stuff.
A bit weird, but...


Regards,
Jean-François


More information about the Digikam-users mailing list