[Digikam-users] xmp and shortcuts

[Jean-François Rabasse]

On Sun, 13 Apr 2014, Photonoxx wrote:

> This thread is a bit old, but I start to use Lightroom cause I change my 
> Camera (now with Fuji X-E2) and my usual Developper (Nikon Capture NX) 
> accept only Nikon Raw or tif/jpg.
>
> So I encounter the same "issue".
>
> I clearly understand the way Digikam and Darktable use. It's clearly more 
> logical but Lightroom is a well know software, and cut interoperability 
> with it may bring people to stop using Digikam or Darktable.
>
> My side, I don't make copy of XMP file, but I've made symbolic link. I 
> don't know if there's compatibility of windows/linux symbolic links, but 
> since I use Digikam on Linux and Lightroom in Virtualbox with a shared SMB 
> drive where photos are, this way works well.
>
> I just have to refresh metadata in LR and Digikam before starting using 
> them (Lightroom detect XMP has changed, Digikam don't but to validate 
> import, manual acceptance is needed anyway).
>
> After that, I don't know which behavior should have Digikam for writing 
> sidecars... But it would be great if interoperability become possible 
> easily.

Hello Nicolas,

Yes, an old thread, but a really older problem :-)

As for me, I strongly doubt interoperability could be achieved, one day
or the other, because choosing a sidecar naming convention is not a 
technical issue but a philosophical issue.

When Adobe created XMP, in the early 2000s, they explained their logic
and stated on the difference between an image and an image file.
An image is supposed to be a representation, a scene, a photo shot...
This image can then be stored on computers drives under different
formats, raw, png, jpg, others.
And the major idea is that it remains the same image, using such or
such raster file format doesn't change the visual.
(Except in cas a user would use the same base name for different
pictures stored under different formats, e.g. Image1.png would have
nothing common with Image1.jpg. Hum, dangerous IMHO)

Given that, as XMP metadata contain technical shot information
(the xmp:exif schema) and documentation information, title, keywords,
author, credits, etc., it make sense to consider that XMP data
relates to the image, not to an image format or another.
And this leads to a convention where the image is referenced by a
base name, Image, the different possible formats by the name and
a specific extension, Image.png, Image.jpg, and the XMP data by a
unique file, Image.xmp, valid for all the formats.

But this logic isn't shared by every one, and other people may
consider that XMP metadata relates to such or such file format.
And this leads to convention ImageName + format extension + .xmp

No one is right or wrong (even if human beings often consider they
are right and the others are wrong :-)
It's more a matter of taste and also of usage.
For my personal usage, I don't have XMP metadata that could be
associated to one image format and not the other. So, I will
prefer the Adobe logic, rather than having to edit a title, or
set tags or keywords, on several files (png version, jpeg version
et al.), with the same values. I tag images, not files formats.

But I think this problem will never converge to something more or
less standard, mostly because sidecar files isn't a standard per se.

Interoperability would also concern images viewers, not just images
editors. Most viewers, KDE Gwenview, Gnome EOG, etc., display XMP
data along with the image, but they all use the XMP data embedded
in the currently displayed image file.
No use of sidecar at all.

So, probably, interoperability means storing XMP data into the
files, and keeping sidecars as working tools for such or such
application.

And adapting such or such naming convention for sidecars to the
requirements of such or such application will still remain a kind
of hacks and tweaks. Probably forever...

Regards,
Jean-François