Shockingly bad keyword assignment on import

[newmikeman]

Please excuse me if I sound a bit grumpy; I feel that I have pretty much wasted a week on this software. I still believe that PSU has some positive potential but I'm rapidly losing faith.
All my photos have delimited keywords embedded, like Europe|France|Limousin|Haute Vienne|Limoges and People|Family|Sarah.
Before importing any I ensured that all my top level labels (what PSU calls Categories) existed in PSU; I edited a couple of names and added a few extra ones.
I ensured that the Preference to read delimited keywords was set along with the correct delimiter (pipe).
I did have a go at importing my whole hierarchy of keywords/labels but that proved impossible because everything arrived under a single category called Lightroom Keywords.
I imported a small subset of my photos, 1867 images representing several different cameras and time periods.
Many of the keywords in the image files matched up successfully. Many did not. The result is a mess and I have lost a lot of vital information, or I would have if I left it like this.
In the Places, we managed to get some more levels in parts but the lowest level label names sometimes got truncated.
Miscellaneous seems to be the most important category and contains a hundred labels (including Places and People!), some with sub-labels. All of these should have matched with categories and labels and/or caused labels to be created in the proper hierarchy.

I have read a lot of posts on this forum and it seems that keywords/labels are a complete minefield.
There appear to be some arcane and specialist approaches to labeling (or key-wording), many of which require me to change the metadata in my image files to meet someone's recommendation.
Some of the posts seem to have a semi-religious tone, indicating that if I have chosen a slightly unconventional hierarchy then I'd better look out or the Lightroom police will be onto me. I came here to escape from the Lightroom police. I mainly came here to be able to relax and quit maintaining the photo library software that I wrote 15 years ago.
Please someone suggest something I can read to help me understand.

thanks
Mike

[Mike Buckley]

You mentioned only some of your Preference settings. I recommend that you post screen shots of your Read and Write Synchronize Settings.

[newmikeman]

Thanks. Good idea.

Capture2.PNG (28.15 KiB) Viewed 7010 times

Capture1.PNG (33.54 KiB) Viewed 7010 times

Regards, Mike

[Mike Buckley]

Your settings don't explain for me the cause of your issues, though perhaps others will pick up on something I didn't notice.

I have a workaround you could try: Start with a new database. Change the Read Keywords processing to "Do Not Convert Keywords to Catalog Labels." Import some files. Return the Read Keywords processing to "Merge Keywords with Existing Catalog Labels." Select the imported images, right click them and select "Metadata" then "Read Metadata from File."

On separate subjects:

I recommend enabling the Write setting, "Write Catalog Data to IDimager ICS scheme." Doing so stores all information in the file such as versioning, relationship catalog labels, Portfolios and other information that is proprietary to Supreme. Once that information is stored, you can easily recreate your catalog even if your catalog becomes corrupted and if you have no current backup.

I don't understand why you are using both Lightroom Hierarchical Keywords and Delimited Keywords. If you use only Lightroom Hierarchical Keywords, the hierarchical characteristic will be maintained. That and the simple keywords (the keyword at the lowest level of each hierarchy) will both be written to the file. Moreover, you have chosen the pipe character as your delimiter. The Lightroom Hierarchical Keywords use the same delimiter, so there is nothing gained as far as I can see by using Delimited Keywords.

[jstartin]

In another thread you said "I'm migrating from a different software but using the so-called Lightroom keyword import feature of PSU". As well as posting your PSU settings it might be helpful if you provided more information about the software you used to write keywords originally and what it was set up to do, and some example files containing the metadata as presented for import.

[Hert]

You talk about delimited keywords....but when you're a Lightroom user then id like to point out that Lightroom is not using delimited keywords but hierarchical keywords.

[newmikeman]

Thanks for your replies.

Mike,
I did as you suggested and starting from a clean empty database I changed the Read Keywords processing to "Do Not Convert Keywords to Catalog Labels.". Then I imported a small set of 51 images. Nothing has been assigned to any Category.
Next I set the Read Keywords processing to "Merge Keywords with Existing Catalog Labels" and then selected all those I had imported.
Read Metadata from File resulted in most being labelled in the catalog. Now there are 2 photos under Miscellaneous. Under the deepest-nested category (below) the deepest subcategories are not truncated as they had been before.

Capture3.PNG (23.44 KiB) Viewed 6953 times

The 2 labels under "21 Annes Grove" have been truncated as "For" and "Gard", as happened before, not "For the record" and "Garden" as is present in
There are 4 images under Miscellaneous, and the Properties for one has the tags "People|Family|Mike Newman; Places|UK|Bucks|Milton Keynes|Great Linford|21 Annes Grove; Other; Styles|Portrait;". The "Other" tag has been placed under Miscellaneous and all the other hierarchical tags have been matched correctly. There are no other photos (in this batch) referenced by the Miscellaneous|Other tag.
There are some photos with a simple tag such as "UK" which in my system will be mapped to the correct label under Places because in this particular hierarchy "UK" is unique. I feel it would be fair to assume that the same approach would be adopted in any other system, such as PSU.
However it seems that simple tags go into Miscellaneous and the ones containing hierarchical tags with pipe characters are added as new categories if the highest level one is unknown, like "Things" in the above screen shot..

Thanks for your suggestion to set the "Write Catalog Data to IDimager ICS scheme" option; should I also set the corresponding Read setting?

I do not use Lightroom. I am using software that I wrote myself and it records the hierarchical keywords as ordinary tags using the pipe delimiter. I implemented that so as to be able to import to Lightroom but when I got LR I hated it so much that I abandoned it. I left the PSU setting to Read and Write hierarchical keywords as well just in case that was relevant given that I wanted to have a keyword hierarchy in PSU.

jstartin,
I think I have answered your question just above.

IDimager,
Lightroom accepts delimited keywords on import and that is why my software implements them. I'm afraid I never got to grips with XMP.
My expectation of PSU is that it will also accept delimited keywords on import and match them with the hierarchy that will be built up as I import more photos. Once all the photos have been imported I would be able to turn off the Read/Write of delimited keywords as you recommend and rely instead on PSU being able to write XMP hierarchical keywords to my image files. So I'm relying on PSU to import my keywords correctly.

I'm going to carefully examine some of my image files to confirm that there are no incorrectly recorded keywords in there. I've done this before of course but I would like to be sure.

Any more ideas please?

regards, Mike

[Mike Buckley]

The 2 labels under "21 Annes Grove" have been truncated as "For" and "Gard", as happened before, not "For the record" and "Garden"

I can't explain that.

However it seems that simple tags go into Miscellaneous and the ones containing hierarchical tags with pipe characters are added as new categories if the highest level one is unknown, like "Things" in the above screen shot.

Assuming you are referring to simple tags already existing in your files before importing the metadata into PSU, that is by design. That design makes sense to me.

Thanks for your suggestion to set the "Write Catalog Data to IDimager ICS scheme" option; should I also set the corresponding Read setting?

No. Reading the ICS scheme slows down the speed. You would enable that setting only when something has gone very awry with your catalog, requiring you to read the ICS scheme to recreate your catalog. Once you have recreated your catalog, you would then disable that setting until the next time you get in trouble and need to use it again. For the record,I've never had to recreate my catalog or use a backup copy of my catalog. That fact means that I'm that much closer to eventually having to do so.

[vlad]

Hi Mike,

The 2 labels under "21 Annes Grove" have been truncated as "For" and "Gard", as happened before, not "For the record" and "Garden" as is present in

Could you write down which are your tags in one of the images with those labels?

There are 4 images under Miscellaneous, and the Properties for one has the tags "People|Family|Mike Newman; Places|UK|Bucks|Milton Keynes|Great Linford|21 Annes Grove; Other; Styles|Portrait;". The "Other" tag has been placed under Miscellaneous and all the other hierarchical tags have been matched correctly.

As Mike B pointed out, that is by design - and that design makes sense to me as well.

There are some photos with a simple tag such as "UK" which in my system will be mapped to the correct label under Places because in this particular hierarchy "UK" is unique. I feel it would be fair to assume that the same approach would be adopted in any other system, such as PSU.

I understand what you're saying. Please try the following:
- turn off "Read hierarchical keywords"
- delete from the catalog one of the images which was (incorrectly) assigned the Miscellaneous::UK label
- re-import the image

Is the image now assigned the proper Places::Europe::UK label?

My expectation of PSU is that it will also accept delimited keywords on import and match them with the hierarchy that will be built up as I import more photos. Once all the photos have been imported I would be able to turn off the Read/Write of delimited keywords as you recommend and rely instead on PSU being able to write XMP hierarchical keywords to my image files. So I'm relying on PSU to import my keywords correctly.

I think your expectation is reasonable and it could be met by PSU, if you make the necessary adjustments.

[newmikeman]

Thank you all for your patience and for constructive replies.
I've been back to the source system and spend time verifying that it has been exporting the right data as expected, and there were one or two anomalies after all. Having corrected them I have tried all over again from the start with PSU and the result is a great deal better.
I've still got those 4 photos in Miscellaneous, but as you say Mike that is to be expected. 4 out of 1734 isn't bad, and if I extrapolate to the full 20,000 images I should expect about 46 to deal with manually - well within the bounds of reasonableness.
The truncated keywords appear to come from IPTC's standard limiting keywords to 64 characters. I think I may have to bend the standard to accept longer ones with all the hierarchy of names and pipe chars.

Thanks Mike for clarification re the ICS settings.

Now that I've got a pretty solid base I think I'll spend a little time exploring some of PSU's other features and useability.

Regards, Mike

[Mike Buckley]

Nice to know you've gotten past that hurdle. Best of luck moving forward!

[vlad]

I've still got those 4 photos in Miscellaneous

Does any of your Miscellaneous labels duplicate labels somewhere else (e.g., Places::Europe::UK)? If not, then you're probably good to go for the big import. I wouls still advise you to disable the reading of hierarchical keywords until you succesfully import all your images. Good luck and let us know how it goes.

[vlad]

Any update, Mike? Did you successfully import all your 20k pictures?