I dumped Google photos for self hosted immich and it’s been great. Not sure if it has the ability to connect to photo frames but if you already self host anything else it’s not too much work to setup.
This is one of the few things I can get behind with machine learning. Locally run, doesn’t “phone home”, just analyzes faces and categorizes them. That’s it.
It’s really not that heavy. I have Immich - and several other services - running on a 4-core VM with 16GB RAM, running on 7th gen Intel hardware alongside another 4c/16GB VM and several LXC containers. It does the job just fine with more than enough overhead.
Self hosted doesn’t necessarily mean hosted at home. Some of my stuff is for example hosted in a French datacenter for redundancy. At one point it was my only server space since hosting at home wasn’t feasible at that time.
I dumped Google photos for self hosted immich and it’s been great. Not sure if it has the ability to connect to photo frames but if you already self host anything else it’s not too much work to setup.
I went to install immich but it was very heavy… I don’t need AI in a selfhosted Google Photo’s replacement.
a self-hosted* Google Photos* replacement
I’m sorry, but then you are the minority. AI image categorization and face recognition is amazing for finding specific pictures quickly
I wasn’t super thrilled about the idea myself, but with the ability to disable it I figured I’d give it a try.
Personally it’s a really cool feature that adds to the experience.
This is one of the few things I can get behind with machine learning. Locally run, doesn’t “phone home”, just analyzes faces and categorizes them. That’s it.
That’s great when your photos have faces. Mine mostly don’t. Also they don’t come from my phone. So Google is also fairly useless.
It also works on simple objects, not as good as Google yet but good enough for me. I can search for keyboard or bicycle and find pictures that way.
That’s good to know. I’m using digikam, but I’m always interested in options
That’s fair.
Those features can be disabled to ease up on resources.
I got it running on a raspberry pi and have disabled the machine learning parts in the docker compose file. Works great.
It’s really not that heavy. I have Immich - and several other services - running on a 4-core VM with 16GB RAM, running on 7th gen Intel hardware alongside another 4c/16GB VM and several LXC containers. It does the job just fine with more than enough overhead.
I have it running on 4gb of ram. No problems!
Try Nextcloud Memories.
Does that run on top of next cloud or can I run it independently of next cloud?
It’s a Nextcloud app.
deleted by creator
This is the way.
Check out Immich Kiosk as an example for image frames.
The only problem is, my internet’s upload speed is slow as hell.
Self hosted doesn’t necessarily mean hosted at home. Some of my stuff is for example hosted in a French datacenter for redundancy. At one point it was my only server space since hosting at home wasn’t feasible at that time.