Needs two write backs for keywords

[DigPeter]

If I delete the old KW and write the new one, then save, I find that 2 write-backs are needed.  To overcome this I have save after each action (i.e. after deleting the old and after writing the new).  I fear that this suggestion would result, on my system at least, in 2 write-backs being necessary, resulting in the case of a large batch replacement, a long extra delay.

[Mario]

If IMatch needs two write-backs to get your keywords in sync, carefully check the existing keywords in your files.
In 99% of all cases two write-backs after changing keywords is caused by out-of-sync flat/hierarchical keywords in the image.

[DigPeter]

I do not understand that Mario.  A long time ago (either during or shortly after the B test) we had a fairly lengthy discussion about this.  The conclusion I came to was that I had to live it.

[Mario]

I only ever encountered this (just this week!) when a file has different keywords in legacy IPTC keywords and/or XMP:subject and XMP hierarchical keywords.
IMatch then needs one write, reload, another write to make all keywords the same. Only by looking at one of your files in the ECP you can tell.

[DigPeter]

Mario

I do not think that I have any legacy IPTC.  I conducted test with the images in the attached zip. 

Test-01 is the original file
Test-02 is the same file after deleting a keyword and writing another.  In this case save was only done once.  Result: 2 write backs.
Test-03 is Test-01, after the same keyword was and deleted and then saved;  then a new keyword was written and saved.   Result:  only one write back.

[Mario]

None of the files in the ZIP has any keywords. Attached the wrong files?

[DigPeter]

Oh - The originals did but I reduced size to to fit to limit set by the community forum.  Here is another set to which the same comments apply as in in my last post.

[Mario]

Your files contain legacy IPTC keywords. XMP subject (flat keywords) and hierarchical XMP keywords.
The keywords are not in sync as I guessed and hence IMatch needs two iterations to set things straight.

[IPTC]          Keywords                        : Peter Photo, Bourne Valley, Hants, UK, Wild Plants, Eranthis hyemalis, Ranunculaceae
[XMP-dc]        Subject                         : Peter Photo, Bourne Valley, Hants, UK, Wild Plants, Eranthis hyemalis, Ranunculaceae
[XMP-lr]        Hierarchical Subject            : Source|Peter Photo, Where|UK|Hants|Bourne Valley, Subject|Natural History|Wild Plants, Taxa|Flowering Plants|Ranunculaceae|Eranthis hyemalis


Compare the keywords in the org. file with the keywords after the second write-back to see how synchronizing is performed.

[DigPeter]

IPTC and XMP-dc lines look the same to me - what is out of sync? 

[Mario]

Hierarchical Keywords and flat keywords.

[DigPeter]

Right - I begin to understand.  In the Thesaurus  I exclude Source, Where, Subject, Natural History, Taxa & Flowering Plants  from flat keywords.  So there will always be a sync problem unless I can prevent the keywords being written to IPTC.   I do not particularly want keywords in IPTC, so how do I prevent it?

[Mario]

The problem is created by Where|UK|Hants|Bourne Valley I think
I don't know your flattening settings or what you exclude or not.
But you produce two flat keywords from one hierarchical keyword: "Hants" and "Bourne Valley" and IMatch has to map both back to their hierarchical counterpart when re-importing the keywords.
What is the difference after the 2nd write-back? Did you check that?

You can strip all legacy IPTC data from your files via the ECP and the corresponding preset.

[DigPeter]

The problem is created by Where|UK|Hants|Bourne Valley I think
I don't know your flattening settings or what you exclude or not.
But you produce two flat keywords from one hierarchical keyword: "Hants" and "Bourne Valley" and IMatch has to map both back to their hierarchical counterpart when re-importing the keywords.
What is the difference after the 2nd write-back? Did you check that?

Here are the metadata after the 1st and 2nd write backs:
1ST WRITE BACK
[IPTC]          Keywords                        : Peter Photo, Wild Plants, Eranthis hyemalis, Ranunculaceae, NW Hants
[XMP-dc]        Subject                         : Peter Photo, Wild Plants, Eranthis hyemalis, Ranunculaceae, NW Hants
[XMP-lr]        Hierarchical Subject            : Source|Peter Photo, Subject|Natural History|Wild Plants, Taxa|Flowering Plants|Ranunculaceae|Eranthis hyemalis, Where|UK|Hants|NW Hants


2ND WRITE BACK
[IPTC]          Keywords                        : Peter Photo, Wild Plants, Eranthis hyemalis, Ranunculaceae, NW Hants, UK, Hants
[XMP-dc]        Subject                         : Peter Photo, Wild Plants, Eranthis hyemalis, Ranunculaceae, NW Hants, UK, Hants
[XMP-lr]        Hierarchical Subject            : Source|Peter Photo, Subject|Natural History|Wild Plants, Taxa|Flowering Plants|Ranunculaceae|Eranthis hyemalis, Where|UK|Hants|NW Hants

You are correct that the Where keywords are the problem. It would also apply to any other multi node hierarchical keyword.   I want those flat keywords, so there is nothing that I can think of to get round this.

You can strip all legacy IPTC data from your files via the ECP and the corresponding preset.

OK - I have managed to find out how to do that.   Would it be possible to have a metadata preference setting not to write IPTC keywords?

[Mario]

Would it be possible to have a metadata preference setting not to write IPTC keywords?

No.

IMatch, in accordance with Metadata Working Group rules and recommendations, only updates existing legacy IPTC metadata. It by default does not create this old metadata format.
But when a file has legacy IPTC, keywords must be synched. You cannot let rot a different set of IPTC keywords in your files.
Just let IMatch write back twice to sort it out. Seems to work with your files.

[DigPeter]

The vast majority of my image  files are freshly sho tand have keywords written only by IMatch.  Why do they contain IPTC keywords?

OK - I can write back twice, but with a large batch of files, this takes a lot of extra time.  My system is not as fast as yours!

[Mario]

IMatch does only write IPTC keywords when it detects a legacy IPTC record in the file. To determine this it checks about 10 fields of the metadata extracted by ExifTool.
I here assume that you use the Metadata, Metadata2 and per-file metadata settings defaults.

I have so far not seen a file that has no legacy IPTC data but somehow makes IMatch think that it has.
Send me one of your files so I can look at the metadata ExifTool extracts.

[DigPeter]

Mario - thanks or all your help with this little problem.

I have now discovered how all my images have IPTC keywords.  In my workflow, having written written keywords in IMatch, there are NO IPTC keywords.  I then edit the file in LR and export it.  This introduces IPTC to the edited file.

However, as explained in my first post, if I save after deleting the keyword and before writing the new one, only one write back is needed.  So that is what i will continue to do.

[Jingo]

I'm starting to think hierarchical keywording is a mistake and really just introduces more issues than it solves (for most folks).  I for one have a mixed bag of keywords... my keywords from 18 years ago (thumbsplus days) are all flat... simple IPTC stuff.  Somewhere along the way, it was recommended to use the hierarchy to tell orange the fruit from orange the color... so - off I went for new photos.  However, if I really look at my thesaurus and keyword tree.. about 97% of them are perfectly fine as flat keywords... though some provide context (Place->USA->Texas->Austin->Salt Lick BBQ) I probably could get by just fine with Texas, Austin, Salt Lick BBQ all flat.

What is the current recommendation regarding keywords for the non-pro, non-stock agency photographer?  Is there one?

[Mario]

There is nothing to say against flat keywords.

This is why IMatch allows you to separate the organization of your keywords (via the thesaurus) from the actual keywords assigned to your files.
Some users get by with a few dozen keywords. Others need a few hundred. Scientific users work with taxonomies with 10,000+ keywords on 10 or more levels. The same is true for some commercial users, train and plain spotters etc.
People selling stock photos have large and 'fluid' keyword thesauri and often add 50 or more keywords to a file.

In general: Use the simplest system that meets your needs.
You may need only flat keywords. Maybe some with a second or even a third level. Or you use the Thesaurus only to structure your flat keywords for quicker access.

And, not everything has to be mapped to a keyword...
IMatch gives you the powerful categories concept to manage multiple concurrent organization forms for your files outside of regular keywords.
Very quick to change, no update of files required when you reorganize your categories.

[Aubrey]

I have now discovered how all my images have IPTC keywords.  In my workflow, having written written keywords in IMatch, there are NO IPTC keywords.  I then edit the file in LR and export it.  This introduces IPTC to the edited file.

I'm now working with DxO. Your comment explains where all my IPTC keywords are coming from, I used to use LR.

I'm starting to think hierarchical keywording is a mistake and really just introduces more issues than it solves (for most folks).  I for one have a mixed bag of keywords... my keywords from 18 years ago (thumbsplus days) are all flat... simple IPTC stuff. 

I'm almost reaching the same conclusion about keywords. The idea is great but cleaning up and moving is a real headache for me, I think most issues coming from IPTC. Previously I just worked with categories.

In mitigation I think if one is starting fresh then using keywords is definitely the way to go.

I'll persevere for another week!
Aubrey.

[Jingo]

Yeah... I think I've just been plugging along the same way for so long that I haven't stopped to really think about the best way for MY photos... since I don't do a huge amount of keywording per image (max 5) and I don't have a ton of "overlapping" keywords with multiple meanings (ie: Orange or 'Bill') - perhaps it is time to let go of the hierarchy... anyway.. thx for the discussion!

[Mario]

Always thing from "How would I like to be able to find these files?" when you think about keywords or category organization.

Unless you have to adhere to academical standards or client requirements, you are free to choose whatever system suits you. Simple is best.