Tags and stars - interoperability through the workflow

I am sure it’s a horribly bad idea and you can now explain to me WHY I am so wrong, but …

tags and stars should be useable through the entire workflow and/or translated fom one program to the other

Yeah, weird idea. I tag something in shotwell or I mark a good shot with a star in geeqie and once the import is done, those tags and stars are already taken in account in darktable or … whereever. I’m sure this is too “integrated” because everyone who is spending his time on free software is already doing God’s work, everyone had a different approach to tags and stars and coordinating these means that

  • a standard has to be agreed upon
  • that standard has to be applied
  • and a lot of programmers have to spend hundreds of hours on this
  • some programs will probably need a total work-over due to this and … there is only so much time

It probably starts already with the question “three or five stars - what will we do?” Three is enough, really:

***Excellent (very rare - competition or magazine/portfolio material)
**Good Enough (good photography, subject well presented, the internet and my family will like it)
*Maybe Look Again Later (not bad … there’s something, but … maybe I see it four weeks from here?)

What other criteria do you have that I don’t?

I didn’t say this will be easy, neither do I fully understand the problems nor the workload. But we are approaching Christmas and this is my secret wish-list. Being catholic also gives me the right to expect miracles at any given moment … so …

I use 5 Stars:

***** Excellent
****  Good Enough
***   Maybe Look Again Later
**    Crap but interesting for Gimp experiments
*     Crap
      Not rated

I don’t seem to … dissect and analyse my crap as much as you do, but hey! let’s go with five stars, I am not obliged to use them all and it makes you and others happy

we use xmp:rating in darktable, which is not in dublin core, unfortunately. it’s often assumed to be integral type, in dt there are 0-5 stars plus the choice of -1 (rejected). for my limited use cases this is overkill already, and it seems the adobe xmp toolkit also allows floating point values. in the past people convinced me that the range up to 5 stars would be generally agreed upon.

so i’d say there is your standard, and also quite a few programs would likely be able to read and modify the xmp blob. there’s a question about embedding it into the raw file (very bad idea) or how to name the sidecar file. the xmp standard proposes name clashes (not a great idea either).

that said, i’m not disciplined enough to know any more about standards like this, and i don’t have personal uses cases. the xmp format itself i think posed more problems than it solved for dt (external programs overwriting your history stack settings, giving the impression of compatibility/interoperability where there is none etc).

There should be a preference in geeqie to write the DT style xmp files, name.ext.xmp. Then set darktable to look for changed xmps.

I use a similar system to you, except my 1 Star shots get deleted, and my 2 Star shots are usually ‘crap, but keeping for personal reasons’ such as memories. 5 star systems seems pretty standard.