I am thinking of developing a convenient interface (with MS Access, in Windows 11), handling the SQLite from my DT 5.2.0 installation
Goals are tagging (adding new one or bulk modifications) and rating (both stars and colour labels)
Does it sound reasonable or, at least, feasible?
DB structure seems to be quite intuitive, although I don’t have access to any more in-depth documentation that merely exploring the DB with DB Browser.
For instance, in library.db tagged_images have a position field, which I can’t imagine what it means… I will need to add new tags, for sure.
Another option (far more difficult for me, is developing this subsystem in lua
Thanks
It sounds reasonable, but not universally useful. Many of those using DT are Linux, rather than Windows users.
There are a number of web based DB clients that will work with SQLite, these are OS agnostic.
I’m not at all a database man (sql always seemed completely unintuitive to me) but… if something works in Access, is it portable to Libre Office equivalent?
Or is that one of those If I was going there, I wouldn’t start here ideas?
This sounds impractical unless I misunderstood something. How can you tag and rate images in a GUI db without seeing the images?
If tagging is an issue you might want to explore the import / export options for using text files,
Thanks
My main goal is develop a more friendly User Interface than DT tagging module. The new UI will allow to reformulate categories and tag hierarchies in bulk
Additionally, I would use that new UI to star and colour label assigment
I have no document on DT Data Model in database, so I don’t know the meaning of fields like “position” in images and tagged_images tables. And I can imagine this is key
It think I can load several images in a MS Access form (possibly from selected_images table), for refined tag assigment, although not in RAW format.
Modifying the database outside of darktable, especially without a THOROUGH understanding of the databases(there’s more than one) involved is probably a really good recipe for disaster. I would make sure you have good backups before starting.
What does that mean? What isn’t convenient about the darktable rating/tagging/color label interface?
You are saying you want to create a whole new GUI to show images and tag and rate modifying the dt database? That’s a lot of work and you should also consider the xmp. And then only a windows solution?
How about improving the dt tagging module? What do you find deficient in it?
Thanks, is there any documentation available on the data model (more than just table definition)?
I just found 2 db files in DT library.db and data.db
Hi,
Since I had to use an application to manage metadata, I preferred DigiKam, which handles this task perfectly.
The metadata, stored simultaneously in the DigiKam database and in .xmp files, are read by ART.
Can Darktable display metadata generated by an external application?
I have no idea on lua programming
I think I have to create a UI to 1 table in data.db and 3 tables (in fact, a view of few columns in the case of the images table) in library.db, and perform some basic operations.
Unfortunately no. The only resource I know of is reading code.
No problem: it’s written in C.
But, joking aside: if you have some generally useful suggestions, raise a feature request to benefit everyone.
Upps, well, thanks.
So, I better give up before starting. Updating xmp is another difficulty not worth the effort
Yes, entering the lua field would be interesting, but even the “Hello world” script doesn’t work for me
I’ll consider if there is any suggestion for improving the tagging system
no need to give up so fast. If you get comfortable with data structures in SQLite database, then you can do a lot of useful actions - but also it’s easy to shoot yourself into your foot.
Keep in mind there’s no referential integrity between library.db and data.db but especially for tagging you need consistency between those dbs.
And you need to consult the code for not so obvious data like “position” in tagged_images.
All in all once you’re successful you’ll have enough insides to suggest improvements for darktable with possible solutions - quite more valuable as those “I want this like that” requiremnets