Comment by Radim
5 years ago
> We realized soon that it's hard to monetize such [self-hosted product]
Spot on. We iterated on a similar product in this space: "privacy preserving", "self-hosted", "open source" etc. But focused on local AI indexing & search of personal videos and photos [0], rather than backups.
We ultimately shelved VideoNinja because we weren't able to find a sustainable business angle:
* Non-technical people simply don't care (happy locked into Apple / Google).
* Technical people understood the proposition, but are super stingy. Case in point, see the responses in this very thread: "$10 per year max; I can buy a HDD for less!". That's one (cheap) restaurant meal per year.
So I fully understand your decision to go "cloud". Although that immediately takes your product off the table for me personally. I want nothing of mine (of value) in the cloud.
I feel there must be a way to square that circle, the market exists.
Just put a price on it, ffs! Make it extensible with plugins. To gain 100% trust make it open source. I am happy to pay good money of a local, non-leaking AI based tagging software for video and photos.
> To gain 100% trust make it open source.
I think until they've got a customer base and a proven model a happy median is to put the code in escrow and agree to give the source to paid licensees should the project be abandoned/more than x months without updates/whatever.
Very surprised no one has mentioned Synology yet. This has been done. And it's awesome!
I currently have a self-hosted google photos clone and I only paid for the hardware. Highly recommend.
Synology's Moments is ok, but it has issues. Not mobile friendly at all, can only create one shareable link per album, and others can't contribute their pictures to your album. Those are the biggest issues in my experience.
I'm still not satisfied, but photoprism seems to move into the right direction here. Digikam os great of you want everything on a single machine. Shot well has other advantages. None of them have a good solution to immediately and automatically import any photo taken on your phone.
Use apps like PhotoSync and it will upload automatically when photos are taken.
While it unfortunately didn't work in the consumer market, there's a space for video recognition in the business space:
- Scene finding for directors/news channels. AP and other sources have a lot of material but you pretty much literally have to watch the entire video in order to find a good scene.
- Scene finding for the XXX crowd. Very underserved market.
- Scene finding for police/lawyers. While it may seem like the opposite of 'privacy preserving', defense attorneys are literally just swamped with video evidence in an attempt to make them give up. Similarly if you're suing a big company for something as simple as an on the job injury or harassment, and need to prove there's a pattern of harm... they'll give you everything and let you do the work of finding out that there was a pattern of bad behavior.
It's the kind of thing that'd be useful as an open source solution... or failing that having a company which is 100% neutral in operation is also good.
I'm currently using Microsoft for something like this because they're absolution massive and apart from their OpenAI division, they only care that what you process is legal.
> I want nothing of mine (of value) in the cloud.
What's the issue with the cloud if you encrypt client-side? It's off-site backup. Isn't it too risky to have your life's work on a few drives in the same location?
And then after a year of usage it hits the news that they botched the encryption, or that they helpfully back up the encryption key in the cloud too.
I’d pay for this if it could run locally. Not sure what it would take to be sustainable but solving this problem is worth at least $20/month to me.
I think too many technical people have too much of a distrust of the cloud. I, for one, am happy to offload as much as possible to the cloud (except latency-sensitive things like games) and not carry around drives and drives at home.