Deploy Immich vs Ente

Self-hosted photo library vs Ente's encrypted cloud

Deploy Immich vs Ente

Just deployed

/data

Just deployed

/data

Just deployed

/cache

/var/lib/postgresql/data

Deploy and Host self hosted Immich vs Ente (Open-Source Photo Library) on Railway

This template runs Immich 3.3.3 on Railway, with a machine-learning service doing CLIP smart search and face recognition on the server. This comparison is for families weighing Immich against Ente Photos and its encrypted cloud.

About Hosting Immich vs Ente open-source software on Railway (self hosted Immich template)

Immich keeps the originals. Whoever holds the Railway project can open the files, so this is not end-to-end encrypted; Ente is, since its Museum server only ever stores blobs.

Railway handles the wiring: private networking, the Valkey job queue, Postgres credentials and the machine-learning URL all come pre-set.

Why Deploy Immich vs Ente, the Ente Photos alternative on Railway (Railway Free Trial)

Ente keeps the server blind by design, which is the point; Immich keeps the whole library and its search index in your hands. Here you pay for compute instead of a per-storage plan, and the $5 Railway trial is enough to push a real test upload through.

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.

By deploying Immich vs Ente on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.

Railway vs Other Hosting Providers and VPS for Immich vs Ente self hosting

Railway runs all four Immich services, domain and private ML included; elsewhere you assemble them yourself.

ProviderWhat You Get with RailwayWhat You Get with the Other Provider
DigitalOceanDomain, private ML, VectorChord PostgresDroplet, your own compose, TLS and backups
AWSManaged domain and logs, no IAM workEC2, RDS and S3 you assemble by hand
HetznerSame template, a volume you growCheapest compute, you own TLS and updates

Ente self-hosting needs Museum, Postgres and an S3-compatible bucket on either side, and no managed template does that for you.

Common Use Cases for hosted Immich vs Ente

  • Family library with face search: choose Immich. The CPU machine-learning service does face grouping and CLIP indexing on the server, so phones stay light.
  • Privacy-first storage: choose Ente. Museum stores only encrypted blobs in Postgres and the S3 bucket, so whoever runs the box cannot read the photos.
  • Moving off Google Photos: Immich has immich-go and the Immich CLI for Takeout archives; both apps back up the camera roll afterward.
  • Shared albums with relatives: Immich shared albums and partner sharing work per account; Ente public albums run on port 3002.
  • A low-power server: Ente starts around 1 GB RAM and 1 CPU; Immich needs more headroom for the ML models.

Dependencies for Immich vs Ente Docker hosted on Railway

Four services ship here and all 36 variables are defaulted, so your work is mostly reading them. Valkey 9.1.2 runs the BullMQ queue and needs a password. Self-hosting Ente instead adds one more dependency, an S3-compatible bucket.

Deployment Dependencies for Managed Immich vs Ente Service (Photo Library)

  • immich-server v3.3.3, port 2283, volume /data.
  • immich-machine-learning v3.3.3 CPU, private port 3003, volume /cache.
  • Postgres 14 with VectorChord 0.4.3, private, volume attached.
  • Valkey 9.1.2, private, volume /data, password required.

Implementation Details for Immich vs Ente (Using Immich official docker image)

The server runs ghcr.io/immich-app/immich-server:v3.3.3 and pulls DB_HOSTNAME, DB_PASSWORD, REDIS_HOSTNAME and REDIS_PASSWORD from the other services. IMMICH_MACHINE_LEARNING_URL points at the ML private domain on port 3003. Change TZ from Etc/UTC to your zone.

How does Immich vs Ente compare against other self-hosted photo services

Immich vs Ente Photos (Ente Alternative)

  • Where ML runs: Immich does CLIP search, face recognition and duplicate detection on the server. Ente is end-to-end encrypted, so that work runs on your devices instead.
  • Who sees files: the Railway project owner can read /data on Immich. Ente's host sees only encrypted blobs in Postgres and the S3 bucket.
  • Setup: four services and a first signup here, versus Museum, Postgres and an S3 bucket wired by hand for Ente.
  • Ente wins on: a managed cloud with family plans, if you would rather run nothing at all.

Immich vs Google Photos (Google Photos Alternative)

  • Ownership: originals sit on your own volume. Google counts photos against your account storage and sells Google One tiers for more.
  • Polish: Google wins on automatic people and place grouping and on sharing. Immich covers timeline, map, shared albums and partner sharing.
  • Migration: export with Takeout, then pull the files in with immich-go or the Immich CLI.

Immich vs PhotoPrism (PhotoPrism Alternative)

  • Mobile backup: Immich wins here; its apps back up in the background. PhotoPrism's mobile story is weaker.
  • Browsing: PhotoPrism is strong on browsing and AI tagging, though some of that sits in paid membership tiers.
  • Weight: PhotoPrism is a lighter Go app. Immich needs Postgres with VectorChord, Valkey and the ML service, roughly 4 GB RAM.

Immich vs iCloud Photos (iCloud Photos Alternative)

  • Ecosystem: iCloud wins if the whole household is on Apple gear. Immich is cross-platform: iOS, Android and web.
  • Storage: iCloud+ paid tiers, versus a Railway volume billed per GB-month with no per-seat cost.
  • Keys: with standard iCloud, Apple holds the keys. Immich keeps the files in your own project, readable by you.

How to use Immich vs Ente (the OSS Google Photos alternative)?

  1. Open the Railway domain and create the first account; it becomes admin, so do it right away.
  2. Install the Immich mobile app, enter your server URL, and log in with the email and password you just created.
  3. Turn on background backup; the server peaks near 1.5 GB RAM while thumbnailing.
  4. Wait for machine-learning jobs: thumbnails, CLIP embeddings and face detection. On CPU a large library takes hours.
  5. Try smart search and face grouping. In Ente both run on your phone, not the server.
  6. Create a shared album and confirm the link opens for a relative on their phone.
  7. Back up the /data volume and Postgres off-site; nightly dumps land in /data/backups, and a volume is not a backup.

How to self host Immich vs Ente on other VPS Services (Immich self hosting guide)

Clone the Repository

Make an immich-app folder and pull docker-compose.yml and example.env (save it as .env) from the latest Immich GitHub release. For Ente, the quickstart script writes a my-ente folder with compose.yaml and museum.yaml.

Install Dependencies

Docker Engine plus the compose plugin. Ente also needs an S3-compatible bucket, since Museum only stores encrypted blobs. Upstream Immich asks for 6 GB RAM and 2 cores (8 GB and 4 recommended); Ente runs on about 1 GB and 1 CPU.

Configure Environment Variables

Set UPLOAD_LOCATION, DB_DATA_LOCATION, an alphanumeric DB_PASSWORD and TZ in .env. Keep the Postgres data on local SSD, never a network share.

Start the Immich vs Ente Application

Bring the stack up and open port 2283 for Immich or 3000 for Ente's web app.

docker compose up -d

Without SMTP, Ente signup codes show up in the server logs. Railway handles the compose wiring, private networking and VectorChord Postgres for you.

Official Pricing of Immich vs Ente (Immich pricing)

Immich is free under AGPL-3.0, with no paid tier and no hosted cloud of its own. FUTO sells optional one-time supporter licenses, and they unlock nothing extra. In both cases your real cost is infrastructure: Railway compute plus volume storage, or a VPS.

Immich vs Ente cloud vs self hosted comparison (Pricing, features, costs, and more)

Self-hosting trades a subscription for your own labor and storage bills. Here is what each path actually costs you.

Monthly cost of self hosting Immich vs Ente on Railway

Expect roughly $20-40/mo in compute for light family use, about 4 GB RAM across the four services, plus volume storage billed per GB-month as the library grows. Self-hosted Ente on a spare box costs you the box plus S3 storage instead.

System Requirements for Hosting Immich vs Ente on a VPS

Immich's upstream minimum is 6 GB RAM and 2 CPU cores, with 8 GB and 4 cores recommended, and Postgres wants local SSD, never a network share. Ente's Museum server needs about 1 GB RAM and 1 CPU, plus Postgres and an S3-compatible bucket.

Frequently Asked Questions (FAQs)

Is Immich end-to-end encrypted?

No. Anyone with access to the Railway project can read the originals, and the ML service needs that to build face vectors. Pick Ente if whoever hosts the server must not be able to see your files.

Can Ente be self-hosted?

Yes, AGPL-3.0. Museum runs with Postgres and an S3-compatible bucket via Docker Compose, about 1 GB RAM and 1 CPU. Without SMTP, codes appear in the server logs.

Where does face recognition run in each?

Server side for Immich, in the ML container, with results in Postgres via VectorChord. On your devices for Ente, because Museum cannot decrypt.

Which needs more server resources?

Immich. The ML service holds about 1.8 GB with CLIP, face and OCR models; plan roughly 4 GB across the project. A minimal Ente stack fits in about 1 GB.

Can I import from Google Photos?

Yes. Export with Takeout and import with immich-go or the Immich CLI. Ente has its own import path.

Which mobile apps back up in the background?

Both. The Immich app takes your server URL and backs up automatically once enabled, and the Ente apps can point at your self-hosted Museum.


Template Content

More templates in this category

View Template
NEW
RustFS | (Just Updated) S3 Object Store, Public URL and Keys Set
S3-compatible object store, keys and public URL preset, volume kept

SuperSlowSloth
0
View Template
TimescaleDB
A PostgreSQL-based time-series database for analytics on Railway.

Rakit
1
View Template
Garage S3 Storage
Ultra-light S3 server: fast, open-source, plug-and-play.

PROJETOS
8