keywording...

[space]

Hi,

For want of a better term, looking for a way to "break apart" merged keywords.

At some point, probably with IDimager and while learning about these things, went and keyworded a whole heap of images. While doing so, may have enabled the write option with "Merge Catalog labels with keywords". The net result is that I have hundreds if not thousands of images with what I now consider to be "malformed" keywords - things like places.city.New York, Keyword.photographer, Keyword.writer, People.writer.Joe Blogs and whatever.

Short of going through every image with Bridge or ExifTools, is there another way of clearing out all these redundant "malformed" keywords? I'm looking to get rid of words like "Keyword" and "breaking apart" items like places.city.New York into separate keywords. Well, don't need to break them apart as these keywords already exist in the hierarchical keyword structure.

Looking forward to suggestions.

Cheers - Rogan

[jstartin]

To get this clear, do you mean that you have unwanted entries in the XMP keywords tag and showing up in the Details Panel?
If all is well with the labels in the catalog itself then you should be able to correct the XMP by using RightClick/Metadata/Save Metadata to File after setting the appropriate preferences.

[space]

Hi Jim,

I think it goes deeper that this. In working with and then moving on from IDimager to PSu and then a few other apps in between, many of these images have been done over a few times. Just recently have been working with the Keyword tab in Bridge to "clear out" the keywords from heaps of these images.

In response to your query and looking at the File Info in Bridge, the keywording mentioned above seems to be in the IPTC fields as opposed to XMP. Or, are these one and the same thing?

At the moment I'm working through and reworking a lot of my Catalog/Category tags. Not sure if this will help fix up the mess I seem to have created. Let's see how we go on this.

Cheers - Rogan

[tstoddard]

If all is well with the labels in the catalog itself then you should be able to correct the XMP by using RightClick/Metadata/Save Metadata to File after setting the appropriate preferences.

Rogan,

I think the critical part of Jim's response was "setting the appropriate preferences". I made a mess of my keywords when I first started cataloguing my files by writing delimited keywords and also writing lightroom style keywords to my files. I then started using a raw processor that would write keywords to my jpg's when the jpg's were output. That program recognized lightroom keywords and would split them properly into single (undelimited) keywords. I'd end up with individual keywords, which it got by interpreting the lightroom keywords plus the delimited keywords that I already had in the file. So if I had something like your example, "places.city.New York", I would end up with four keywords: "places", "city", "New York", and "places.city.New York". You can imagine what it would look like if I had 3 or 4 delimited keywords to begin with.

I thought I could fix them by just changing my settings to not write delimited keywords and then saving metadata to file again, but that didn't work at first. I think what I had to do was to change the setting for merging keywords to replacing keywords. I'm sorry I'm not at a computer with PSU installed so I can't tell you exactly where that is in the preferences dialog.

Also, it sounds as if you've already figured out how to clear your keywords from your files but I think you could have done that from the details panel also. As long as your labels are configured correctly, it should be safe to clear out all of those keywords and then just do as Jim suggested and save metadata to file.

The other more difficult issue to resolve is changing your label configurations if you have a lot of labels set up incorrectly to begin with. PSU doesn't enable you to act on multiple labels simultaneously so if you have a lot to change it takes forever and is somewhat error prone. I had also made the mistake at some point of setting up a lot of my labels to write parent labels to keywords. This was even harder to fix because I had to go into every single label and change them one at a time. I missed some of them in the process and it took a long time for me to identify the ones I missed. I'm still finding files with keywords in them that I don't really want. I also found that changing the labels and saving metadata to file didn't remove the parent label keywords because I didn't have the merge vs. replace preference set correctly.

Of course, what ever you decide to try, I would recommend that you test it on a smaller set of files before doing it to a large set.

[jstartin]

In response to your query and looking at the File Info in Bridge, the keywording mentioned above seems to be in the IPTC fields as opposed to XMP. Or, are these one and the same thing?

When you label images in PSU the labelling is stored in the catalog database. Synchronization options allow catalog information to be written to the image file (or a sidecar file). When you synchronize with a writable file, PSU will write keywords to both the IPTC block and XMP.

I think the critical part of Jim's response was "setting the appropriate preferences"

That is certainly critical, but so is the condition "if all is well with the catalog itself". For example, does the unwanted "Keyword.photographer" keyword appear because it is a hierarchical catalog label (seen in the Catalog by Categories view of PSU)? If it does then the catalog probably needs to be tidied using PSU's tools to move labels, merge labels etc.

[space]

Hi Tom and Jim,

Appreciate both you guys chiming in on this one.

@ Tom - seems like we've both being trekking through similar territory here - as in working with something thinking that we know what we're doing only to find out that the results proved otherwise. And you are right as is Jim - its the way we've set up the preferences. Since with working with PSu, have "neutered" things to the absolute minimum.

As indicated in my initial post, all I want now are simple keywords - as in "places" "cities" "New York" and completely forgoing "places.city.New work".

@ Jim - as can be surmised from the above, I am running a "hierarchical" keyword structure - this in the "strict sense" of the word "hierarchy" - as in levels, hence "place/location", "country", "state", "city", etc. However, and as Tom suggests, I don't want this "hierarchy" to be presented in dot notation or other delimited form. This is fine for applications that can read and parse out this data but, for the Internet and other searches in general, this is no good.

In following up both your suggestions and taking a closer look at preferences, this what I have set up at the present moment. In the "Read settings" > "Keywords processing" I have "Merge keywords with existing Catalog labels". Everything else remains "unchecked". This setting seems to make sense to me in that any new keywords will be matched up with existing keywords in the Catalog database instead of creating new and possibly potential duplicate keywords. This is my take and, may be this is where my issues lie?

Under the "Write settings", I have this. Under "Keywords processing" > "Do not modify keywords". I've checked the following: "Only update out-of-sync images...", "Store keywords in alphabetic order", "Store synonyms as keywords too", "Include all parent level labels as keywords".

@ Tom. In the above - with this write function, is this where you suggest going over to the "replace keyword" thing?

As for working with one image at a time, what more can I say that hasn't been said here and elsewhere before? This is why I'm using Bridge to try and fix up this mess.

Let me soldier on...

[jstartin]

I realise that one source of confusion in corresponding about this is that "keyword" is so widely used in the context of other DAM software that PSU users tend to forget that Hert uses the term "label" for what is in the catalog. It is essential to differentiate between what is in the catalog, which is what PSU uses for searches etc, and the optional writing out to image/sidecar file to facilitate interchange of metadata with other applications. I was not quite certain whether or not you were using keyword as a synonym for label. Now it is clear you were not. So...

As indicated in my initial post, all I want now are simple keywords - as in "places" "cities" "New York" and completely forgoing "places.city.New work".

@ Jim - as can be surmised from the above, I am running a "hierarchical" keyword structure - this in the "strict sense" of the word "hierarchy" - as in levels, hence "place/location", "country", "state", "city", etc. However, and as Tom suggests, I don't want this "hierarchy" to be presented in dot notation or other delimited form. This is fine for applications that can read and parse out this data but, for the Internet and other searches in general, this is no good.

Under the "Write settings", I have this. Under "Keywords processing" > "Do not modify keywords". I've checked the following: "Only update out-of-sync images...", "Store keywords in alphabetic order", "Store synonyms as keywords too", "Include all parent level labels as keywords".

I suggest trying with
- Select "Replace keywords with Catalog labels" to get a fresh start with all the old, unwanted keywords removed and replaced by keywords corresponding to current catalog labels.
Edit: No reason not to do the replacement across your entire image collection even if some keywords are already OK, unless you can easily isolate the images that do need correction.
- Deselect "Only update out-of-sync images..." while you sort things out, so you can easily rewrite keywords for everything according to current preferences. I don't think that simply changing preferences will make images go out-of-sync even if the keywords would be written differently.
- Deselect "Write delimited keywords" (having this selected gives you the "dots").
- I think you could safely enable "Always write Lightroom hierarchical keywords", and "Write catalog data to IDimager ICS schema" which would allow your hierarchical labels in the PSU catalog to be regenerated from the XMP data if this ever becomes necessary (if normal backup fails), but should not upset other applications (they will not read these tags unless they know how).
- Whether or not to store label synonyms and label parents as keywords is your choice. (With my catalog I always record a location for the shot, a subject for the shot (which might be a place), and usually a style, so writing undelimited label parents as keywords would put these words into the metadata for every photo I have!)

In the "Read settings" > "Keywords processing" I have "Merge keywords with existing Catalog labels". Everything else remains "unchecked". This setting seems to make sense to me in that any new keywords will be matched up with existing keywords in the Catalog database instead of creating new and possibly potential duplicate keywords. This is my take and, may be this is where my issues lie?

The read settings should not really matter unless you are going to read from XMP back into the Catalog. However, I would make the setting "Do not convert keywords to catalog labels" to limit the potential for unwanted keyword entries making their way back into the catalog. I don't really understand your point about new keywords being matched up with existing keywords in the catalog. Are you using/allowing another application to write keywords?

Edit: It occurs to me that, compared with what seems like prudent normal-procedure, you have the keyword processing settings for write and read reversed.

[tstoddard]

In the "Read settings" > "Keywords processing" I have "Merge keywords with existing Catalog labels".

Tom. In the above - with this write function, is this where you suggest going over to the "replace keyword" thing?

Rogan, I'm not on a computer with PSU installed but I think that the "Merge keywords with existing Catalog Labels" setting is in the "Write settings" under "Keyword Processing", where you say you have "Do not modify keywords" selected. I think there are three chooses for that setting. One of them is to "Replace existing keywords with labels". My suggestion was to change that to Replace if you want to overwrite existing keywords with new ones based on your new label configuration.

Also, I came to the conclusion that if I "Always write Lightroom hierarchical keywords" that I don't need to "Include all parent level labels as keywords". The parent level keywords will be in the Lightroom keyword hierarchy so you probably don't need them in your keyword field as well. It's possible you might have a reason to do that anyway but I have not found it to be necessary.

I haven't used synonyms yet so I don't have that option set and I see no benefit to storing my keywords in alphabetical order so I don't have that set. Also, Jim is correct to say that the "Read settings" should not make much difference to you, but they can have a slight impact on performance and there is no reason to have most of them selected unless you are in the process of rebuilding your catalog based on the data that is stored in your image/sidecar files so you might want to take some care as to which ones of those settings you select. I don't remember them all off of the top of my head so I can't tell you what I do and don't have selected.

If you're using other programs to modify your keywords, that could necessitate changing your settings to accommodate for that. It is generally not a good idea to use more than one program to modify metadata. Getting two programs to work together is extremely difficult and sometimes makes it impossible to avoid causing errors or inconsistencies in your catalog and file metadata.

[space]

Jim and Tom, thanks once again for chiming in here.

@ Tom - your comments more or less coincide with Jim's comments. I've now implemented the suggestions put forward by both of you and will start with these shortly.

Edit: It occurs to me that, compared with what seems like prudent normal-procedure, you have the keyword processing settings for write and read reversed.

@ Jim - You may well be right on this - my dyslexia cutting in again, even now I have to shake my writing hand to remind me which way is right and which way is left. Luckily I'm right handed.

It is generally not a good idea to use more than one program to modify metadata.

@ Tom - you are spot on here. I think this is something that's contributed to this issue in the first place. Have typically used Adobe's Bridge to do most of this work. At one point, in the change between older versions and CS5, Adobe made some changes and this did some "damage" somewhere. Can't remember the issue now but, its was quite significant.

@ SYNONYMS: In the scheme of things and when it comes to putting content out on to the Internet, these things are crucial if not critical. Use synonyms mainly for spelling differences and the like. Examples like colour {color}, Montreal {Montréal}, for difficult to spell people names like Ketil Bjornstad {Ketil Bjørnstad}, in abbreviations like Oxfordshire {OXON}, New South Wales {NSW}, for country codes like China {CN} and Canada {CAN}. Then, we have something like this: Hong Kong SAR {Hong Kong, HK, HKSAR, Hong Kong Special Administrative Region}. In another instance and while working with Getty Images, there their software would parse out names like this, "BJORNSTAD, Ketil". So, this gets added in as well. While seemingly complex (and rather deep, I'm afraid), this a "do once and apply to all" thing. In much the same light, what we take for granted may not be obvious to others and what is obvious to others is something we may overlook. A matter of covering one's bases.

@ other synonyms - "keywords" and "labels" - in this environment they seem mutually interchangeable. A bit like "locations" and "places". While a local Starbucks may be a "location" for a photo shoot, it is usually a "place" on some street in some city in some state/province/county in some country. Now, are these all locations or places? (just a test of wit - no need to answer).

@ Lightroom. Don't use and haven't used this app since 3.xx or something. I now use Capture One for all my RAW image processing. While not the most intuitive app on the block, by far prefer the quality of its image output. My decision in this regard was recently vindicated when when it came to processing my Fuji X-E1 images. While I still use Photoshop (CS6) - mainly to "host" my Nik Software apps - my decision to "wean" myself off Adobe products has also been vindicated. While hiring a mechanical back hoe is one thing, renting (or, in this case, subscribing to) a Swiss army knife for everyday tasks is quite another. As for anything "cloud", don't get me started.

Let me get back to work and give this set up a whirl. If there's anything further to report, will get back here.

Cheers - Rogan

[tstoddard]

Rogan,

I have one quick note in response to yours. In case you are under the misconception that I was under at one time, setting PSU to write lightroom hierarchical keywords isn't necessarily done only to provide compatibility with Lightroom. Hert makes that clear in this topic, which you might want to read if you haven't already done so.

[space]

Hi Tom,

Thanks for the link. Just doing a "Sync all out-of-sync images now. 14 hours on and we are a 62.1%. Let's see what happens after this.

Cheers - Rogan