DupeZappa vs Immich: Desktop Toolkit vs Mobile Photo Server
Immich backs up phones to a self-hosted Google Photos-style library. DupeZappa clusters faces on existing Windows folders with local ONNX and ties to dedup and search.
Immich is a self-hosted photo server for mobile backup, partner sharing, and ML face search in a Google Photos-like web app. DupeZappa Face Recognition runs on your Windows PC against existing folder trees with local ONNX models. Pick the tool that matches where your photos actually live.
Full comparison page with feature matrix and verdict →
Keep Immich if your household phones backup to it.
Where Immich still wins
| Strength | Why it matters |
|---|---|
| Mobile backup | iOS and Android apps upload camera rolls to your server |
| Google Photos-like UX | Timeline browsing and search for household libraries |
| Partner sharing | Shared accounts without consumer cloud lock-in |
| Active ML roadmap | Server-side face features evolve with the project |
Immich answers: "Back up every phone photo to our own server and browse together." It does not answer: "Cluster faces on messy Windows/NAS folders without moving files into a server library."
Where DupeZappa extends the workflow
| Module | Problem it solves |
|---|---|
| Face Recognition | Local YuNet + AuraFace ONNX on existing folder trees |
| Smart Tags | Named people as virtual collections without moving files |
| Universal Search | tag:"Name" across indexed drives |
| Duplicate Scanner | Find duplicate photos of the same person in different folders |
The Face Recognition feature page is the direct overlap with Immich's face workflow. Everything else is why some users install DupeZappa on desktop archives instead of re-uploading every export to the server.
Complementary workflow
- Immich: automatic phone backup and household gallery for new photos.
- DupeZappa: face clusters on desktop exports, NAS copies, and old archives that never sync to the server.
Faces find people; duplicate scanner acts on burst shots and copies with preview-first undo in the same operation journal.
Naming and search workflow
- Index photo roots in DupeZappa.
- Run face detection; review anonymous clusters in Face People.
- Name high-confidence clusters; merge splits of the same person.
- Search with
tag:"Grandma Rose"in Universal Search. - Run Duplicate Scanner on the heaviest photo subtree.
Read Organize Family Photos by Face Offline on Windows for ONNX setup and threshold tuning.
vs PhotoPrism and other alternatives
| Approach | Tradeoff |
|---|---|
| PhotoPrism | Mature DAM; albums, places, web gallery |
| Immich | Mobile backup first; partner sharing |
| DupeZappa | Windows desktop; folder-native; tied to dedup + search |
Choose Immich when phone backup to your own server is the primary job. Choose DupeZappa when desktop folder libraries need faces and cleanup without standing up a gallery server.
Privacy default
Face detection and embeddings run locally on your PC. No face data leaves your machine through DupeZappa's built-in path. Immich keeps photos on your server; both avoid consumer cloud upload for core workflows.
Pricing
| Immich | DupeZappa | |
|---|---|---|
| Price | Open source + your server cost | One-time purchase (early access) |
| Primary job | Mobile backup + household gallery | Folder-native faces + cleanup toolkit |
| File location | Library on server | Files stay on disk paths |
Fair read: if mobile backup is the whole job, Immich is the category leader. If desktop folder cleanup matters more, DupeZappa fits without Docker.