The project you are describing sounds close to my goals, with the possible exception of long-term storage: I've been planning to develop everything so that data is easily retrievable even if the software doesn't work. Metadata in text files, for example, image storage in well-known formats, folder structure that is understandable to humans, etc. This is for family archives: I'm scanning photos that are a hundred years old and I feel like an idiot when cataloging them using software that some company will obsolete in a couple of years.
Incidentally, there is another niche which is largely unaddressed: easy organizing of photos into sets with various access tags and sharing these with friends and family in a secure and easy way. It's kind-of doable today if you don't care about privacy, security, and longevity, but it's still annoyingly difficult to use, especially for less computer-literate family members. It seems that all designers of photo software only keep photos from their latest vacation, wipe their entire collection with each device upgrade, and want sharing links to be valid for at most a month or so.
One thing which could prevent me from using your software is lack of control: I got burned several times already and I'm considering my options carefully. It might be better to roll my own rather than start depending on something again.
File storage (if you opt for PhotoStructure doing automatic organization) is configurable (with a default of $library/yyyy/yyyy-mm-dd/$original-filename).
All metadata extracted or changed is saved in standard-format sidecars next to the original, or in the image itself. The sidecar format and whether to use sidecars for edits are both configurable.
The database is a SQLite DB with a straightforward schema (although as I add features, each version has had db migrations).
If you try PhotoStructure out, and you find or think of anything else I can do to set things up to be more friendly to long-term storage, I'd love to hear your ideas.
(Sharing is coming in a future release, both via any hierarchical tag, or by manually-selected images).
There is still the issue of control: I am worried about the lack of a "Pricing" page, not because I do not want to pay (quite the opposite), but because I am worried about the sustainability of your business. And about what will happen when you sell it to Google, or otherwise abandon it.
FWIW I've got a bunch of both scanned images and older JPGs that I inadvertently stripped all metadata from (if you use jpegtran to losslessly rotate, make sure you use `-copy`!), so PhotoStructure infers captured-at time from filenames or enclosing folder names if it's missing from the EXIF headers, and it can extract at least a reasonable year, month, and day from the path.
There's also PhotoStructure for Servers which can run under Docker, or as a standard node app (ala `git clone $url; photostructure/photostructure`).
My guess is that the same people, when given a native app, complain about the subscription pricing or the price tag.
And I happily pay for all of the quality native software I can find. One of the most exciting things is finding a well-crafted native application with a clear business model.
I'm really hoping Serif, for example, moves into the Photo organization space. I've already purchased all of their current apps even though I don't really need them all that much; probably 70% of the reason I bought them was just to support what they're doing.
https://photostructure.com/about/introducing-photostructure/
It's currently in closed beta. I'm letting people in first-come-first-served roughly monthly, as I release new versions. It's free while in beta, in exchange for your feedback. Here's what's coming soon:
https://photostructure.com/about/release-notes/
Note that PhotoStructure doesn't have any editing tools other than fixing rotation. It does let you easily open the current image in other apps, though.