Live data from Hacker News

Immich v2.0.0 – First stable release

github.com

11–20 of 165 posts

Re: Immich v2.0.0 – First stable release

#11
post #6

I haven't heard of this app before, so I looked around their site and docs. I was mildly interested in trying it out until I saw the requirements: "A system with at least 4GB of RAM and 2 CPU cores.", but recommended 6 GB of RAM. Why does an image storage solution needs so much RAM?

Because it's not just an "image storage solution". A thumb drive would be an image storage solution. If you're indexing, making geo queries, serving over the network, categorize, transcode video and everything else that's needed to create a google photos competitor, you're going to need the hardware to back it up.

Re: Immich v2.0.0 – First stable release

#12
post #6

I haven't heard of this app before, so I looked around their site and docs. I was mildly interested in trying it out until I saw the requirements: "A system with at least 4GB of RAM and 2 CPU cores.", but recommended 6 GB of RAM. Why does an image storage solution needs so much RAM?

Because it is written in Node, it uses several other softwares (i.e. Postgresql), it runs image/video processing on the fly (transcoding to various formats depending on what you upload and who views), it does face/object recognition running a local model and a few other nice features that yeah, require more power. It's not a static HTML of your photos.

Re: Immich v2.0.0 – First stable release

#13
post #2

Immich is excellent. Especially the beta timeline introduced in the last few Android app releases has leveled Immich up from 'pretty decent' to a genuine replacement for Google Photos for me.

Yep, with the new timeline they fixed my biggest issue with Immich which was the local app won't even show your local photos if there was no connection to the server. Now it works, you just don't get others photos if they are not cached already but that's absolutely expected.

Re: Immich v2.0.0 – First stable release

#14
I tried to use the embedding stuff a year ago but the results were lackluster, even with the larger embedding model.

With the new multimodal LLMs it seems a better approach might be to get a multimodal LLM to describe the image and list keywords, and then just use the included Meilisearch.

That said, I see they list some models I haven't tried, so perhaps time to try again.

Re: Immich v2.0.0 – First stable release

#15
post #7

I appreciate the Cursed Knowledge section on their website https://immich.app/cursed-knowledge > Cursed knowledge we have learned as a result of building Immich that we wish we never knew

Reading it I see this

  50 extra packages are cursed

  There is a user in the JavaScript community who goes around adding "backwards compatibility" to projects. They do this by adding 50 extra package dependencies to your project, which are maintained by them.

which bring to this user: Jordan Harband https://github.com/sponsors/ljharb Does anyone know what they actually mean with that cursed knowledge point? And what's the "backwards compatibility" that Jordan also boasts in his GH profile?

Re: Immich v2.0.0 – First stable release

#16
post #7

I appreciate the Cursed Knowledge section on their website https://immich.app/cursed-knowledge > Cursed knowledge we have learned as a result of building Immich that we wish we never knew

Agree, I wish every project would have that!

I still think the conclusion on "setTimeout is cursed"[0] is faulty:

> The setTimeout method in JavaScript is cursed when used with small values because the implementation may or may not actually wait the specified time.

The issue to me seems that performance.now()[1] returns the timestamp in milliseconds and will therefor round up/down. So 1ms errors are just within its tolerance.

[0]: https://github.com/immich-app/immich/pull/20655

[1]: https://developer.mozilla.org/en-US/docs/Web/API/Performance...

Re: Immich v2.0.0 – First stable release

#20
post #7

I appreciate the Cursed Knowledge section on their website https://immich.app/cursed-knowledge > Cursed knowledge we have learned as a result of building Immich that we wish we never knew

Reading it I see this 50 extra packages are cursed There is a user in the JavaScript community who goes around adding "backwards compatibility" to projects. They do this by adding 50 extra package dependencies to your project, which are maintained by them. which bring to this user: Jordan Harband https://github.com/sponsors/ljharb Does anyone know what they actually mean with that cursed knowledge point? And what's t…

Seems like this thread answers your question https://news.ycombinator.com/item?id=37604373
Post reply on HN