While I do organize my images in darktable, there are reasons why a reasonable file system structure makes sense for me:
darktable only handles images, but I keep movies and audio recordings (all raw material I generate) in the particular event folders.
darktable refuses to read certain files, e.g. the panorama jpgs of my phone. While technically they count as broken, an adaption of darktables jpg reader was refused, but other software reads them without issue, such that I need to find them outside of darktable.
I can remember more easily which files are already in the catalog and such folders still have to be added.
darktables film rolls use the folder name.
It is easier to copy an event on file system level.
I leave all my pictures wherever they fall. I search, view and tag with XnView MP which has the best search function Iāve seen yet. I wrote my own hierarchical structure (text file) which can be found here:
I do not trust black box āsystemā, from experience, I prefer shelves and libraries rather than mechanical jukeboxes. Indexes can still be rebuilt, modified and reorganized.
Originals/2021/20211221/20211221_214600_00.DNG for raw files dating back to 2001. My first few years I used jobcodes in the day directory names, then I scratched it because it didnāt work (for me). I can shoot a myriad of different things in a day and descriptional metadata fits better in the original file or a sidecar.
Why the _00 before .DNG you ask? If there are two or more images taken the same second I need a two digit sequence. I donāt know if Rapid Photo Downloader can do that? Some cameras actually record SubSecTimeOriginal that seems tempting. But, as all cameras Iāve had historically donāt support it that would result in a mix of sequencing and SubSecTimeOriginal. Also, a future camera you buy might not support it, so I would really advice against using it for naming files. Iāve used exiftool forever and I do this:
The first round picks up still images (and videos) with DateTimeOriginal. The second round picks up video files that does not have DateTimeOriginal and uses MediaCreateDate instead.
Yes, year is in three places and month in two. Somehow I concluded this was a wise choice, perhaps I wouldnāt do it again. However, with this naming Iām never unsure what year or month Iām currently in.
Raw files exported to TIFF go in a directory structure parallel to Originals called Derivatives they are in full resolution. From these I create Delivery files.
Thatās a good workaround for the problem of keeping file names unique when sticking with a strictly date-time-only filename format. But why not use sequence numbers? The site you link to in your post suggests exactly that.
A job sequence for the entire import or day uses more digits. For me Iād need four. Also, if you do some culling after renaming youāll have large gaps in your sequence. Not aesthetically pleasing. But yes, I have images which only have _02 but not _00 or _01 because _00 and _01 have been culled post renaming.
That could have a practical explanation. The popular tools amongst professionals (like Lightroom) donāt support āsub second sequencingā. I donāt think C1 has it either. I could be wrong. Also, job sequencing works well. But, of course, I think my method is better.