I organize them by year and then ‘photo event’, so yyyy/yyyymmdd-Event. The latter can be a single day (birthday) o multi-day thing (long weekend at the mountains). For anything beyond that, I use tags and a database to filter the images (digiKam and darktable).
I would say you need to find the largest set of info that is unambiguous for every posible image (date+event for me), and use that for the folder structure on disk. The idea is to get a sane file location on disk, where using more information like tags would give you headaches when a picture belongs to multiple categories.
I don’t know if it is a goot practice, but that’s what I do:
Folder structure
I like flat structures, therefore all photos belonging to an event (session, vacation, event, …) will get into folders such as yyyy[-mm[-dd]]-Event_name. If the events span several days, dd is omitted, the same holds for the monts. If multiple-day-events are passing a month barrier, the month when it starts is used, the same holds for the year. And, it’s always the belonging which counts. Example: It was 2017-12-31-New_years_eve, even for the photographs on first of january, as we have been with friends and the reason was the party on 2017-12-31. The 1st and 2nd of January were still belonging to the new years eve party.
Tags
I currently have the following standard hierarchies:
People|Family|Name
People|Friends|Name
People|Work|Name
People|Models|Name
Places|Country/City/whatever
Projects|Year[|Month[|Day]]|Project
Projects|Project (if they are not bound to a special date)
Events|Event
Subjects|Subject
Analog|Here comes a hierarchy of analog metadata such as lab, film, scanner, …
There are also bazillions of legacy tags and hierarchies, but as they do hardly harm, I never spent the time to tidy up.
As the Events hierarchy (and also projects to some extent) doubles with the “film rolls” or folder hierarchy, I tend to use these tags not very consistently. But I always try to improve .
So I try to have every photo in one of the tags Project / Event / Memo
But those single snapshots I take with my smartphone while wandering around, are no project, it’s no event, and it’s not just a memo. So it gets the tag ‘Misc’.
And as ‘Misc’ is just another name for ‘I-don’t-know-where-to-put-it’, this category gets huge and uncontrollable.
Oh, I forgot the crucial part, which initially was the reason I started to answer. For every month I have a folder yyyy-mm-monthname which receives all photos where a dedicated folder would not make sense. It does not matter which camera was used to decide the folder, smartphone photos from a vacation go into the vacation folder. All photos are tagged regarding the same scheme, at least, people are tagged. Images without people should have at least a subject tag, but often I am too lazy.
Folders may include subfolders with names of other photographers. Vacation pictures from my wife will e.g. go into such a folder.
As all my smartphone shots are automatically uploaded into a nextcloud, I recently started to hardlink the synchronized files in the respective folders instead of copying such that I do not duplicate the files. However, not sure if this is a good idea as the files are managed by git-annex.
@chris : do you also have a substructure for the ‘Events’ tags, or is it just a flat list analogous to your folders (Events|2017-12-31-New_years_eve)?
My list of events is quite long (as I have photos since the last millennium), so I tried to create a substructure, but I don’t know yet, if I should structure the events-tags by year or by topic…
a) Events|yyyy|yyyy-mm-dd_title
b) Events|topic1|yyyy-mm-dd_title
YYYY … year
event-name … something like “wedding-jane-joe”
NNNNNN … my unique 6-digit ID
MM-DD … start date of the content in that folder
description-of-content … anything short that will help me navigate the folders in an event
CAM … abbreviation of the camera used
The filenames are just the 6-digit ID, a dash and the number from the original filename or the negative.
The 6-digit ID is a continuation from assigning a number to every physical filmroll, it is basically just a running number assigned to a group of images. This requires some management of the numbers but it frees me from date-related naming schemes and makes searching for a group of tightly related images a blast.
This way every image has a totally unique number across the whole archive, every folder with images in it can stand on it’s own and the year+eventname folders above are just to bundle things together.
Tagging includes rough geolocation (words, never GPS data), names of people and objects visible in the image and a event-id (e.g. WEDDINGJANEJOE)
Yes, I do have such a structure, but I don’t know if it is useful. I have mixed feelings. The theory is, that tags could be combined in searches, which is unfortunately a bit cumbersome in darktable. However, separating keywords therefore makes sense. Therefore, I have event|birthday|2015|Name rather than an entire string. However, this is hard to remember and a flat structure may make more sense, having event, birthday, year and name as separate tags. However, this would interfere with the names of the subjects, therefore the hierarchy kind of makes sense.
In each event’s directory I have subdirectories for each type of source image, e.g., NEFs, JPEGs, MOVs. The top directory of the event contains 800x600 renditions from the source, or what I call “proof jpegs”. With my hack software rawproc, I can drag a proof JPEG to the rawproc window and it’ll open the source, usually a NEF, and apply the processing that rendered the proof JPEG. I can then modify that processing to make another rendition…
The only images that don’t end up in that tree are “technical” images, those taken as part of some testing or construction.
Like @chris, I like a flat structure, but I like it even flatter. I use RPD to import photos, and I pay no attention to any package’s facilities (filmroll, database, or whatever) for cataloging images. That way, my organization has no dependencies on what package(s) I use to work with the files. My file system is my catalog.
My naming convention for directories is yyyy-mm-dd_jobcode_camera
e.g., 2015-10-10_Haliburton-Sculpture-Forest_D7000
If a jobcode is something that may come up again, I can tell RPD to remember it, so if I visit the same area multiple times it’s easy to keep my terminology consistent.
My naming convention for files is yyyymmdd-hhmmss-seq-filename.ext
e.g., 20151010-165544-2562-DSC_0097.nef
So I can easily find things by subject matter or what camera I used, or even the file name from the camera (which might be useful if the SD card hasn’t been reformatted) with a simple ls command.
Like most here, I structure my photos into yyyy-mm $eventname directories. My own wrinkle is that I have a “default event” for each year, accidentally called yyyy $hometown, where all pictures go that were not associated with a special event.
Only exports are organized thus. Raw files go into unstructured daily directories in a separate directory tree.
For file names I generally use $originalname_$rating_$developer, which gives me an easy file-based way of searching for highly-rated pictures (*_4*), or pictures edited in Darktable (*_DT*). I keep the original file name as a prefix because (a) that makes finding the corresponding raw easier, and (b) it tells me the camera model, in a crude way.
My only real problem with the year/event scheme is that sometimes I’ll have images of other things in a given event, e.g., the pictures I took of mangled packages for refund negotiations before/after someone’s birthday party, my OCD about taking unrelated artistic images in the middle of an unrelated event - Squirrel!!!, so on…
Once I’ve rsynced to the backup computer, it becomes harder to segregate those, so I need some more discipline up-front…
I classify images by year, and place and/or topic. The topic may be an event. The squirrel picture would then to the directory for animals taken that year.
I am using digikam to index all my photos (using also face detection)
My choice of directory tree is as follows using exiftool :
Manufacturer / Camera / YYYY / YYYY-MM-DD /
Files are renamed on the fly like that:
YYYY-MM-DD_HH-MM-SS_Manufacturer_Camera_OriginalFileName.ext
no duplicates
very simple. i upload all the RAW files onto two different HDD. they RAw files are kept by their original dates. the exported Jpeg files are saved on two separate HDD there i save them by years. inside each year folder I have the photo folders in chronological order, numbered and a name given to them in order to ease the search.
I don’t structure my pictures by file or directory name, it’s too limited to search, or likely to be modified.
I also avoid modifying my image files by metadata (it also simplifies backups)
My pictures (raws and developed) are in more or less arbitrary directories (why name directories by date, when the system notes the creation date of each directory), like books to borrow, not directly accessible to the public in a library, whose number is only useful to the storekeepers (or to the machines…) to easily find a certain book. The readers who are looking for a book, and the people who establish the indexes, directories and catalogs are not concerned by this system.
To manage all this, I haven’t found better than kphotoalbum.
I totally agree, that’s why I started using metadata + tags to structure my images:
In the first place I use existing metadata to navigate through my photo-archive (datetime, lat/lon).
Then I tag all people with Person|Firstname Lastname
So I already have ‘who’, ‘when’ and ‘where’.
But ‘what’ and ‘why’ are still missing.
According to chris I will tag all images without people with a subject tag (for the ‘what’).
And according to the DAM Book 3.0 a way to structure the ‘why’ is the degree of intent:
‘where I parked my car’ → tag ‘Memo’
'memory images of friends, life events, business documents → tag ‘Event’
‘make a statement, express an emotion, art for arts sake’ → tag ‘Project’
So as mentioned above my current problem is how to structure my ‘Event’ tags.
At the moment, I don’t think I want to structure the tags with ‘date-hierarchies’ (like ‘event|birthday|2015|Name’), because it’s redundant.
But is it useful to find common Event-types for all images (like ‘Event|birthday|John’)?
Or do I just use the verbal description (Event|Johns first birthday), which would be a quite long unstructured list with hundreds of events?
.