# Replycant - self-hosted image storage — retained record

> Retained reddit record from reddit:selfhosted.

Canonical: https://highsignal.app/data/records/3bd709d7b95f9595

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:selfhosted)  published 2026-11-07 12:06 ingested 2026-11-08 02:37

[original source ↗](https://www.reddit.com/r/selfhosted/comments/1xq4uhp/replycant_selfhosted_image_storage/)

## retained text



```
Hi all! I've been working on an open source, self-hosted image management solution for the last 1.5 years. It should be seen as an alternative to iCloud/Google photos/Immich. In august I released the app publicly on github and the iOS app store, and it's stable for daily use.  
  
  
Replycant as of today consists of:  
\- An iOS app on the app store (search for replycant)  
\- A server which is started using docker compose  
\- An electron desktop app for linux/mac/windows  
\- CLI tools for bulk image import from a computer  
  
  
Now, you may ask yourself "why should I run replycant instead of immich or some other open solution"? There are a couple of answers to this:  
  
  
Better data durability - for me personally, the most important thing about an image store solution is that it must never, ever, loose any data. In order to guarantee this, multiple servers are needed which replicate between each other. A hard drive crash should not mean that you loose anything. In order to achieve this, the database in replycant is based on Git and Git LFS, which makes it easy to replicate to multiple sites. Multi-server replication in the app is not implemented yet, but will come soon. As of today, backing up to a new site is still easy though by using a "git clone" followed by periodic "git pull" to sync the latest changes.  
  
  
Better privacy - all data (both metadata in the database as well as photos/videos) are encrypted before leaving the source device. No clear text encryption keys are ever stored on the server. This means that if the server were compromised, no data would be readable. This is important in case of a burglary, or if the server is hacked. However, the real importance comes when combined with the previous point, multi site replication. A simple server such a raspberry pi can be placed in an untrusted location and back up all of your data to a place outside of your home.  
  
  
Simplicity - no configuration of the server or your firewall is needed, a simple "docker compose up" is all configuration you need to start the server. The clients connect using tailscale which traverses firewalls without hassle. Simply put: reach your library outside your home over 5G without any additional configuration.  
  
  
Performance - the goal of the project is to make the app feel like all photos are stored locally on your device. I run the server on a Raspberry Pi 5 with a 2.5" mechanical USB drive which yields excellent performance.  
  
  
Oh, and before you ask: no, there's no Android version yet. I'm working on one.  
And another thing before you ask: yes, the project is implemented using cursor. But no, I don't consider it "vibe-coded". All of the architecture, API data types etc stems 100% from me, but the implementation is done using AI.  
  
  
The project is open source and available at https://github.com/mr-andreas/replycant. The license is GPL2.  
  
  
I'd love to hear your thoughts! :DHi all! I've been working on an open source, self-hosted image management solution for the last 1.5 years. It should be seen as an alternative to iCloud/Google photos/Immich. In august I released the app publicly on github and the iOS app store, and it's stable for daily use.  
  
  
Replycant as of today consists of:  
\- An iOS app on the app store (search for replycant)  
\- A server which is started using docker compose  
\- An electron desktop app for linux/mac/windows  
\- CLI tools for bulk image import from a computer  
  
  
Now, you may ask yourself "why should I run replycant instead of immich or some other open solution"? There are a couple of answers to this:  
  
  
Better data durability - for me personally, the most important thing about an image store solution is that it must never, ever, loose any data. In order to guarantee this, multiple servers are needed which replicate between each other. A hard drive crash should not mean that you loose anything. In order to achieve this, the database in replycant is based on Git and Git LFS, which makes it easy to replicate to multiple sites. Multi-server replication in the app is not implemented yet, but will come soon. As of today, backing up to a new site is still easy though by using a "git clone" followed by periodic "git pull" to sync the latest changes.  
  
  
Better privacy - all data (both metadata in the database as well as photos/videos) are encrypted before leaving the source device. No clear text encryption keys are ever stored on the server. This means that if the server were compromised, no data would be readable. This is important in case of a burglary, or if the server is hacked. However, the real importance comes when combined with the previous point, multi site replication. A simple server such a raspberry pi can be placed in an untrusted location and back up all of your data to a place outside of your home.  
  
  
Simplicity - no configuration of the server or your firewall is needed, a simple "docker compose up" is all configuration you need to start the server. The clients connect using tailscale which traverses firewalls without hassle. Simply put: reach your library outside your home over 5G without any additional configuration.  
  
  
Performance - the goal of the project is to make the app feel like all photos are stored locally on your device. I run the server on a Raspberry Pi 5 with a 2.5" mechanical USB drive which yields excellent performance.  
  
  
Oh, and before you ask: no, there's no Android version yet. I'm working on one.  
And another thing before you ask: yes, the project is implemented using cursor. But no, I don't consider it "vibe-coded". All of the architecture, API data types etc stems 100% from me, but the implementation is done using AI.  
  
  
The project is open source and available at https://github.com/mr-andreas/replycant. The license is GPL2.  
  
  
I'd love to hear your thoughts! :D
```



document fetched 2026-11-08 00:26 · canonical https://reddit.com/r/selfhosted/comments/1xq4uhp/replycant_selfhosted_image_storage
