Comment by istingray
5 years ago
Yes, Ente needs to have self-hosting on their roadmap or I won't support it.
> What? This was extremely random and out of place with the rest of your comment.
Edited to say "open source options may suck today"
Thanks for giving me the chance to explain. My comment here may give more context: https://news.ycombinator.com/item?id=28321460
I've tried NextCloud, even 1-click hosted by a third party. For all the power, it's not built with me in mind, it seems to treat my photos like files/data, not like photos. I want to pay money for that extra oomph, for algorithms, searchability (about $10/month for my photos seems about right), and I want to pay money so I don't have to pay with my time. Is there something I can buy that's on top of NextCloud?
There are many people like you willing to spend money on a good solution that gets the job done who have no interest in self hosting and reviewing the source.
After experiencing the ease of Google photos, any basic file management system to store photos is a downgrade after that.
If ente can figure out how to do the extras (search, face matching) without invading privacy (not even sure how possible this is) I can see this being valuable to the people who want to de-google and maybe even de-apple.
> not built with me in mind
Fairo
> on top of NextCloud
Not that I know of. Could have a look through the nextcloud marketplace for something or another. Tbh, I don't see any open platforms having Google/Apple photos kind of functionality for at least a little bit. Google and Apple trained their algorithms on the people using their free tiers for years. Google especially had access to so much information on the user using Google Photos that it was able to build the algorithms it has today. For an open platform to have this functionality, it would need to wait for an open source model/algorithm to exist, else it would need to build it itself by using user data (no E2EE then).
Unless Google open sourced whatever models it uses in Google Photos today, don't expect this level of searchability yet. Actually even if Google did, it would probably be so tied to Google user information and be incompatible with E2EE.
How to create an open source training set without surrendering my data? Like Numerai but instead of a hedge fund it's photo data: https://numer.ai/