so, one of the discussions in VR this morning turned to how we structure the folder of commissions we've got. the key component in ours hasn't really changed since 2017/2018, but something may have to at some point.
OneDrive/Pictures/Furry/My Commissions/
One subfolder per artist. Acts as source of truth for everything including artist provided; WIPs and full resolution images all go here. Stuff preferably keeps the artist provided filename, even if it means creating a subfolder from them naming multiple things dex.png
iCloud Photo Library
I want to have everything ready to go on phone as well (also those dark few years of using an iPad as main machine). Artist folder with one album per artist, also a separate folder. This only stores webres images so we can't make a mistake on phone.
OneDrive/git/dexthedragon.co.uk/_data/dex.json
This acts the source of truth about metadata, alt text, tags, other people involved, and the like, resizing all images within the main OneDrive folder, with a list of exceptions for images not to touch. Used to build the website, based on MD5 hash - built for speed, not security (no idea what happens if there's ever a hash collision). Built on a collection of scripts that get the MD5, save a resized copy to 1200x1200 max, as well as handling other format conversions and thumbnail generations.
Problems
This setup mostly works, but it does run into the issue that we're slowly trying to move away from OneDrive for other things as we've moved away from Windows. For the moment, this still works, but if we do ever fix up the desktop Linux machine, the only way that can programatically access OneDrive is (as far as I know) Samba to the NAS, which isn't the fastest. Still, I think the NAS would be fine for this, there are some other uses where it might be more of a problem.
Additionally, I've wondered about switching the source of truth on metadata to Obsidian, for easier access to info without going through 8000 lines of JSON - for doing things like linking the source concept for that commission or bits of data that might be useful to me but don't need to be on the website (e.g. outfit here was inspired by this past commission) - but I feel like the things that would make it useful for person readable metadata would make it more awkward for scripting the website; I do not particularly feel like writing a bunch of regexes to get data back out of Markdown. And having them as two separate sources of truth risks them getting out of sync.
There's also an argument to be made for putting the alt text in iCloud as well as a caption, given that our Masto app of choice will now auto fill in alt text based on caption
curious how other people handle it
this is overkill for the vast majority of people, but it also makes it very difficult to lose something permanently given the amount of redundancy and backups
so curious what everyone else does