https://blog.photostructure.com/introducing-photostructure/
Disclaimer: I'm the founder.
How do you manage the "forever" part with a cloud service that could go away?
How is this not just "dropbox, but only for photos"?
It's software that you run, on your hardware. Currently there's a desktop installer for win/mac/linux (so people can try it out easily on their laptop or whatever), but I'm building a multiarch Docker image that runs on a droplet/ec2 micro/NAS/raspberry pi.
I run it for my family off of a rpi 4 sitting in an external 10tb USB drive, plugged into my router.
The external drive gets grabbed in case of fire/earthquake, but also gets backed up to backblaze for offsite backup.
> How is this not just "dropbox, but only for photos"?
Dropbox is a file synchronization service.
* You view your files by folder
* Files can only live in one folder
* There's no automatic organization of your files
PhotoStructure is designed to make browsing millions of photos and videos fun and easy, and effortless.
* Best-of-class metadata extraction and inference
* Automatic organization and tagging of your photos and videos based on that metadata
* Hierarchical tagging, with random "tastes" of what's in that tag (or child tags)
* Automatic, best-of-class "asset variation" merging (so when you have a RAW and JPEG version of an image, along with a Apple Photos resized preview, and a Google Photos takeout, they all are considered to be variants of the same asset).
* Cross-platform, cross-file-system support (so if you plug in an external hard drive and it automatically mounts to a different path, the files on that volume are considered equivalent).
* All viewed through a webapp and image delivery system that's built to be delightful and responsive even on low-powered servers and low-powered mobile devices
PhotoStructure runs on a computer you own, or in the cloud, on a computer you rent. That could be a laptop you have, or a home server, your NAS, or a digitalocean droplet.
Your photos need to be available to that computer when they are imported into your library, either by being on a local hard drive, or mounted from a network fileshare.
Once your photos and videos are imported, the library can be moved or copied to other computers (say, rsync'ed to your droplet). Original photos and videos aren't required.
The PhotoStructure interface is via html/css. It runs a web server that is only available to localhost by default, but there are tools (like localtunnel and trycloudflare) that lets you access your library from any device that has internet access. I'm still building out multi-user support, but a library "owner" has full r/w access, and a "visitor" may see predetermined album contents
I realize that makes me sound crazy, but after years of reflecting on software, privacy, data ownership, and decentralization, I have not been able to avoid the conclusion that Stallman was basically right.
If your data lives in a cluster with a bunch of other people's, it's a tempting target, both to legitimate businesses and to outright theft. It's only a matter of time before someone gets it, as the constant parade of breaches and data sharing scandals show, and once it's out there you can't undo it.
I have speculated that open source services that are easily self-hosted (start the installer, enter a domain name, and pick a cloud provider) might be a passable answer to the desire to have both accessibility, shareability, and independence.
Funding their development is trickier, obviously.
They very easily could be, if the service wanted to write it's contract of adhesion to allow the user meaningful control over the scope of how the service can use the photos.
But conventional wisdom believes there is less money in that route, and nobody in "tech" is willing to settle for less than maximum exploitation of assets.
Practically, it's not. See my concern about security beaches.
Even if the service has a sane ownership policy right now, eventually someone will either change the service's policies or shut the service down.