"The keyword search filter could have a way of searching for the exact term, and the possibility of having several exact terms"
Agreed there should be an "exact match" operator for all fields in the Library Filter bar's Text search and in smart collections. This is an embarrassing oversight in LR's design. For keywords, you can get exact match using the Library Filter bar's Metadata browser with one or more Keyword columns, e.g.
Click the option menu indicated to switch the keyword listing between hierarchical and flat. This example will filter all photos containing both "John Rolfe Ellis" and "Karyn Hunt Ellis" (boolean "and"). Selecting more than one keyword in a column will do a boolean "or" of those keywords.
You can save the current filter as a preset, making it easier to call up in the future.
Unfortunately, the user interface is poorly designed and very klutzy to use for large keyword lists.
A handy shortcut for searching for photos containing one keyword is to hover the mouse to the right of the keyword in the Keyword List panel and click the right-arrow that magically appears:
This will open the Metadata browser with the keyword selected in a Keyword column.
Another handy shortcut: To find a keyword in a large keyword list, use the Keyword List search box:
If you're often searching for multiple keywords (e.g. photos containing k1 and k2 and...), consider the Filter By Keyword command of my Any Tag plugin.
Edit Smart collection allows text to be searched to create a dynamically filtered image collections.
The options are:
contains
contains all
contains words
does not contain
starts with
ends with
From a UX point of view it is not immediately obvious how the first three contains variations are different. You would have to look up help or experiment to figure out what you thought it meant to confirm that it did what you were thinking.
However neither of the "contains" options successfully solves the problem if you are searching for say:
Peter Thomas and Thomas Smith, who might be two different people that you want in two different collections.
For some reason contains all works for one of the collections but for the other collection it includes both Thomas'. Weird. Had to use a does not contain in only one of them to make this work. The asymmetry doesn't make sense either.
What's the business impact? Potential privacy breach if you don't check that the filters are working properly.
The suggestion would be to have a verbatim match which should be called "Exact Match".
1: Contains is an 'or' query. That means that 'Contains Peter Thomas' will find images of either Peter or Thomas.
2: Contains all is an 'and' query. 'Contains all Peter Thomas' will only find images of both Peter and Thomas, but that includes Peter Drake and Thomas Smith.
3: Contains words is a query on full words only. 'Contains Peter' will also find images of Ben Peterson. 'Contains words Peter' will not find these images.
4: If you only want to find images of Thomas Peter, not the combination Peter Drake and Thomas Smith and also not Thomas Peterson, then the trick is to use 'Ends with Thomas Peter'.
I know that many of the experts (including many on this forum) don't like this option, but I have found it useful (especially with people names such as in your example) that share parts of names (e.g. Thomas in your example), but I am starting to use surrogate names.
For example: Keyword name = "Peter-Thomas" (note "dash" between Peter and Thomas), Synonym = "Peter Thomas", uncheck "export this keyword", leave checked "export synonym".
Same thing for "Thomas-Smith"
Now you can filter or set a smart collection that says "Contains All Peter-Thomas Thomas-Smith" and you won't get Fred Thomas or Peter Jones included or even "Peter Thomas Smith"
This technique is obviously a kludge or back door work around but it does work and it's unlikely that Adobe will make any changes to the keywording section of LrC to remedy this long standing problem.
"From a UX point of view it is not immediately obvious how the first three contains variations are different. You would have to look up help or experiment to figure out what you thought it meant to confirm that it did what you were thinking."
I agree that the naming and semantics are a complete mess.
Thanks for merging, John. It is incredible (and mind boggling) that this has been going on for 9 years. Here's my conclusion having read this thread....
I would concur with William and Patrick (from 5 years ago) that from a software point of view this isn't a rocket science change.
But also from a user experience point of view somebody at Adobe has to own this problem. Photographers expect to be able to have a tool that can search their own professional library intelligently. Sure you can make machine learning models that can visually find a toaster, but that isn't what photographers need - they label based on the type of work that specialise in. It needs to be efficient and easily used. Not all photographers no regex syntax :-)
Maybe the issue is "Classic" and that internally they've moved on to the next mobile thing.
I've landed here because I was trying to solve a problem where you have 100 faces in 1500+ images that you have sort. The Face detection is a good starting point, but the failure in keywords just makes this clumsy.
Great I can recognise faces but I can't robustly search for them and file them efficiently.
UX design is caring about the little things and understanding the business context of the user base. Adobe needs to remember that people who use their products are typically running a business (or are part of a business). They invest in tools like Lightroom and Photoshop (and Muse :-p) to keep their business running. Not considering the little things and the workflows that business face every day wastes time (and money).
Somebody at Adobe could use XD to start mocking up a better interface and talk to the software team about the technical hurdles to implement it. You could start with two interns....
I too came across from Aperture and have looked at Final Cut Pro / Motion and Premiere / After Effects. So I have both an Apple and Adobe perspective. I'm also a software developer and UX designer. There are times when I look at Adobe products and just get the sense that somebody who knows UNIX has influenced a particular user interface feature. There's definitely power to do things but nobody hasn't put in the effort to remove the technical clutter and make the product enjoyable to use.
Count how many times you have to reach out for a help guide to figure out a feature. The contains, contains all, contains word pop up labelling is indicative of the underlying issue.
It's not about making everything simple, it's more about designing the experience to be intuitive and providing the professional power when it is needed. Design it for the intended audience.
Fixing keywords shouldn't be difficult. String search technology is well established in 2020 especially if we are dealing with an sqlite back end.
I nominate that the monetary component of my annual subscription that was paying for Muse development should be funnelled into fixing keywords / search in Lightroom Classic.
Thanks for this. You beautifully worded what I mostly feel as "I can't )(*&*(@#&$! believe that in 2020 I still can't (*(*#@$!@ search in a decent )*&*(&#$! way".
Face Detection speeds up the process of organising photos of multiple people.
The most intuitive label for face detection is using the person's name.
However, Smart Collections Any Searchable Text options fails to perfectly match the name because there are no options that understand word order. Names where first names and surnames have similar root words are easily confused.
The end result is you have to start making exception condition rules which also means being on top of every name in the list. Then you have to go through every collection to make sure there are no mistakes - of which there were.
Just tried this on 100 people and the face detection speeds things up, but every gain in time was lost to trying to fix the broken text search.
PS. Don't merge this. It has to been seen by those who own the (newer) Face Detection feature.
The lack of exact match for keywords is frustrating. But a much better workaround surfaced recently using Starts With and Ends With:
This matches the keyword "John Smith" but not "Smith John" and not "John Smithson". It's not absolutely perfect -- it would accidentally match "John Smith John Smith", but I think in practice there wouldn't be any such examples.
(Up until a couple years ago, Starts With was broken and couldn't be used in this way.)
Responses
John_R_Ellis
Champion
•
5.5K Messages
•
97.3K Points
a year ago
Agreed there should be an "exact match" operator for all fields in the Library Filter bar's Text search and in smart collections. This is an embarrassing oversight in LR's design.
For keywords, you can get exact match using the Library Filter bar's Metadata browser with one or more Keyword columns, e.g.
Click the option menu indicated to switch the keyword listing between hierarchical and flat. This example will filter all photos containing both "John Rolfe Ellis" and "Karyn Hunt Ellis" (boolean "and"). Selecting more than one keyword in a column will do a boolean "or" of those keywords.
You can save the current filter as a preset, making it easier to call up in the future.
Unfortunately, the user interface is poorly designed and very klutzy to use for large keyword lists.
A handy shortcut for searching for photos containing one keyword is to hover the mouse to the right of the keyword in the Keyword List panel and click the right-arrow that magically appears:
This will open the Metadata browser with the keyword selected in a Keyword column.
Another handy shortcut: To find a keyword in a large keyword list, use the Keyword List search box:
If you're often searching for multiple keywords (e.g. photos containing k1 and k2 and...), consider the Filter By Keyword command of my Any Tag plugin.
3
ddphoto
33 Messages
•
530 Points
2 months ago
Edit Smart collection allows text to be searched to create a dynamically filtered image collections.
The options are:
From a UX point of view it is not immediately obvious how the first three contains variations are different. You would have to look up help or experiment to figure out what you thought it meant to confirm that it did what you were thinking.
However neither of the "contains" options successfully solves the problem if you are searching for say:
Peter Thomas and Thomas Smith, who might be two different people that you want in two different collections.
For some reason contains all works for one of the collections but for the other collection it includes both Thomas'. Weird. Had to use a does not contain in only one of them to make this work. The asymmetry doesn't make sense either.
What's the business impact? Potential privacy breach if you don't check that the filters are working properly.
The suggestion would be to have a verbatim match which should be called "Exact Match".
Clarity is important in the labelling.
10
0
ddphoto
33 Messages
•
530 Points
2 months ago
Face Detection speeds up the process of organising photos of multiple people.
The most intuitive label for face detection is using the person's name.
However, Smart Collections Any Searchable Text options fails to perfectly match the name because there are no options that understand word order. Names where first names and surnames have similar root words are easily confused.
The end result is you have to start making exception condition rules which also means being on top of every name in the list. Then you have to go through every collection to make sure there are no mistakes - of which there were.
Just tried this on 100 people and the face detection speeds things up, but every gain in time was lost to trying to fix the broken text search.
PS. Don't merge this. It has to been seen by those who own the (newer) Face Detection feature.
1
0