Apple Photos vs SnapByFace: When the Free People Album Isn't Enough
Apple Photos builds a People album for free and never uploads your faces. Here is where it stops — and what you actually gain by adding a separate local index.
Published 2026-09-28 5 min read
- comparison
- macos
- apple photos
If you are on a Mac, you already have face recognition. Apple Photos builds a People album automatically, lets you name people, and asks you to confirm matches as it goes. It is free, it is deeply integrated, and — importantly — the analysis happens on your device; Apple does not upload your faces to do it.
So the honest starting point is: most people should just use it. This page is about the specific cases where it is not enough, which mostly come down to three things: how you search, what happens to video, and whether you are only ever on Apple hardware.
:::note Written September 2026. Apple’s photo features change with macOS releases; check Apple’s current documentation for the version you run. :::
The short answer
- Stay with Apple Photos if your library is photos, lives in Apple’s ecosystem, and you are happy browsing the People album it builds for you.
- Add a separate index if you need to search with a reference photo, need to find people inside video, work across macOS and Windows, or want results as a list you can export.
What Apple Photos does well
- It costs nothing and needs no setup. The People album appears on its own; you name a few faces and it improves.
- Analysis on device. Face recognition runs locally, not in a data centre.
- Integrated everywhere. Same library on Mac, iPhone and iPad, with Memories, collections and scene search alongside it.
- Native performance. It is built into the OS and tuned for Apple silicon; browsing a large library is genuinely fast.
If those cover what you need, a second tool is redundancy, not an upgrade.
Where it stops
You cannot search with a photo of a person
Apple Photos decides who is in your library, groups them, and lets you browse those groups. If someone has not been grouped and named, there is no way to hand it a photo and say “find this person”. There is no reference-photo search.
SnapByFace is built the other way around: one reference photo per person, then search. For archives where you care about a handful of specific people — a client, a speaker, one family member across decades of scans — that is a more direct route than waiting for clustering to propose them.
Video is not searched for people
Footage is stored and Memories will happily use it, but the People album is built from photos. If the question is “which clip does she appear in, and at what time”, Apple Photos does not answer it.
SnapByFace checks video about once per second, compares what it sees against your reference photo, and merges consecutive appearances into time ranges — “02:14–02:34” — so results point at a moment you can jump to.
It is Apple-only
The library, and everything built on top of it, exists inside Apple’s ecosystem. Working on Windows, or handing an indexed archive to a colleague on Windows, means leaving it behind. SnapByFace runs on macOS 12+ and Windows 10+ with the same workflow on both.
Results are for browsing, not export
Apple Photos shows people so you can look at them. SnapByFace lets you narrow a search with a similarity threshold (0.45 by default, 0.30–0.80) and export the result set to CSV — paths, scores, and video timestamps — which matters when the output is a deliverable rather than a browse.
One library, one opinion about it
Apple Photos organises media into its own library structure. SnapByFace is strictly read-only: it reads your folders, never renames, moves or re-encodes anything, and keeps its index in ~/.snapbyface/. Your folder structure stays exactly as you left it.
Side by side
| Apple Photos | SnapByFace | |
|---|---|---|
| Cost | Included with macOS | One-time licence |
| Platforms | Apple devices | macOS 12+, Windows 10+ |
| Where analysis runs | On device | On device |
| How people are found | Automatic grouping → you name them | One reference photo per person |
| Can you search with a photo | No | Yes |
| Video | Stored; not searched for people | Searched by face, results as time ranges |
| Library organisation | Photos library structure | Read-only, folders untouched |
| Similarity control | — | 0.30–0.80, default 0.45 |
| Export results | — | CSV with paths, scores, timestamps |
| Photo management | Full (albums, edits, Memories) | None |
| Needs an account | Optional iCloud Photos | No |
A note on iCloud
Face recognition itself stays on your device, but if you enable iCloud Photos your library is uploaded to Apple’s servers. That is a separate decision from face recognition, and a reasonable one for many people — just be clear that they are two different things. SnapByFace has no equivalent option, because there is nothing to sync.
What SnapByFace does not do
- No photo library. It is not a replacement for Photos. No albums, editing, Memories, or iCloud sync.
- No automatic grouping of everyone. It finds the people you give it, and no one else.
- No phone or tablet app. One machine, the one it is installed on.
- No free tier beyond the trial.
The sensible combination
Keep Apple Photos as your library. Add a local index over the material the People album does not reach: footage, project folders that stay out of your main library, an archive on an external drive, or the scans of a relative who is not in your photos at all.
The free trial runs a complete local search against your own folders and shows up to 5 matches per person and 20 per search. During the trial you can also start up to 100 analyses in total.