Also the docker images now have a major tag v2 that should keep your installation up-to-date on the stable branch.
Finally it will be easier to search my vast catalog of memes.
That’s what I’m waiting for as well. Online meme services are failing.
Do I need to update the IMMICH_VERSION tag in my .env file to v2?
It’s what I did. Or you could keep it on a fixed version.
Have they added the ability to rotate your photos yet?
I don’t see an ability to rotate in the web app. It’s been available for a long time in the mobile app though.
It’s been available… “Kind of”
It creates a copy of your photo and rotates that. Leaving the original. Which I guess I can see being useful sometimes but it’s very annoying in practice last I checked it.
The OCR job is running rather slow though. It does about 3 photos per second with 8 jobs concurrency on a Ryzen 7900.
Im undecided whether being slow is or is not a problem, simply because it’s one of those things that you do once to the entire library and then it’s just half a dozen fotos a day
Oh it’s not a problem on ongoing basis. Just the initial scan is gonna take forever here. 150K photos…
The solution to that is to delete 130k photos :)
I started culling mine years ago, still wading though
Is it the ml service? I an thinking to deploy on the much more powerful desktop and temporarily point the server to it.
😆
Thats just 14h tho, not too bad.
My i3 7100 is about to get a workout
Edit: did 15,000 photos overnight so not bad at all
Laughs in i5-4590
Edit: Oh God I’m getting about 1 image every 2 seconds, 1 image at a time. Looks like I’m gonna be here for the next 14 hours.
Edit 2: it’s been 9 hours. I’m halfway done.
Update to my i7-2600k, it’s taking 6 seconds per image
Yours is faster haha. It’s the 7100T which is the 35w low power model that the Lenovo mini PCs came with. The only advantage is it has decent video decode hardware so it can handle media transcoding better.
I5-2450M for the win
Xeon X5670 😆
Used to have a server as a desktop dual socket where I did upgrade the CPUs to that exact model for like 50€ total
i7-2600k here…
What kind of processing is happening really under the hood, is there an image preprocess, lang detection, script detection, orientation, etc?
I find that the existing search also kind of does ocr
Somewhat. I’m guessing this would be proper full OCR, perhaps even selectable. Still scanning.
Still crutching on containers?
What would you prefer and why? Bare metal?
Gotta chime in with a +1 for bare metal too…
Ideally yea. Rpm package with various architectures. 90% of other software is able to do it. I only use docker for immich and this would remove a dependency for me.
Well…Docker is my only dependency.
Why though? When you run multiple services on your machine, conflicts between these can cause headaches. Updating them is also often not trivial on bare metal. Let alone migrating to a newly set up machine (I suppose you could automate that with something like ansible)
I run jellyfin, plex, mattermost, peertube, minecraft, all under one host. Never had anything conflict and just upgrade the rpms. Some of them use the same postgres database. Thats how I prefer to do things. I will use docker if absolutely forced, I know how, understand its easier to deploy etc. but I would rather it be the choice of the user instead of the software. Or if not rpm let me get in the source and run it.
(…) would rather it be the choice of the user (…)
And that’s totally okay.
Stupid
How so?
Containers are better than bare metal in every way
Containers are better than any other option, so of course they’re being used! I’m gunna use containers even harder!
There’s always one of these comments and it’s always incredibly similar.
Always and forever!
