Author Topic: Keyword handling  (Read 2566 times)

0 Members and 1 Guest are viewing this topic.

David Dyer-Bennet

  • Member
  • **
  • Posts: 21
    • View Profile
Keyword handling
« on: 2014-08-10 19:28:18 »
I've got 172k thumbnails in my database, and a bit short of 100k keyword references so far.  Many of them are of some historical interest for people in the science fiction community. I delve into the old portions of the archive, as well as the new ones, fairly frequently.

For various reasons, like TP not being able to write to certain file formats, and some others, quite a few of the keyword references aren't written into the IPTC fields in the files.  Sometimes this is for file formats that TP can *now* write into, but couldn't back when I applied the keywords.  Sometimes, I believe, it's because some ways of applying keywords over the years (including today) just update the database, they don't write into the file, even for file formats TP can write into.

(Side note: I really wish it were a lot harder to create that situation accidentally!)

Now, I need to avoid having vital data that exists *only* in the TP database.  In addition to my basic paranoia, and my fear of "lock-in", there's an important reason: when I process an image and save a jpeg for the web, for example, I want the jpeg to include the keywords assigned to the image!  *ALL* of them, the ones in the database as well as the ones in the image file.  Now, there's no way for an external program to know what's in the TP database; to achieve this, I think I have to make sure that the image files contain the keywords (as IPTC keywords).

(By the way, this business of not being able to write into certain file formats -- it is of course unavoidable sometimes for a while, when new file formats are released, and for some file formats it may in practice be permanent.  There is of course a well-known, essentially standard, technique for working around this, widely understood by all serious photo software: side-car XMP files.  TP needs to be able to handle sidecar XMP files!)

So, there are three things for moving keywords around that I need:

1)  Make sure all keywords in database are in image file (or sidecar).  This would look up the keywords in the database, and then check each one in the image file, and if it's not there add it.  

2)  Make sure all keywords in image file are in database.  This exists now, it's done when you make a new thumbnail.

3) Do both -- assemble all keywords from both sources, and then make sure they're recorded in both places.

(An automated process to keep them in sync would be desirable perhaps -- except deciding when to allow TP to write into image files is something the user needs control of, so the manual process is necessary anyway.  People change policies, etc.)

As it is, I'm feeling trapped, and I hate that.  I have lots of valuable keyword information that only exists in the TP database, and I have no other way to access it.