Cluster, then merge
Faces are grouped into people. When the same name appears on more than one cluster, the app can merge those people instead of leaving duplicates.
Local grouping
One person. Not twenty clusters.
Faces are clustered into people, fragments that share a name are merged, and unnamed people can be reviewed and named in bulk.
- Precision-first
- Merge fragments
- No cloud
Keeping people together
A face library should not fragment.
Grouping is the part after detection: deciding which faces are the same person, merging duplicates that arrived in separate passes, and giving you a clear place to name the people the app does not know.
Unknown people, in bulk
The Unknown People view collects unnamed clusters so you can name several at once rather than opening each card individually.
A correction surface
The People page is the place to review names and merge obvious fragments. The app proposes; you decide.
Detect faces
The scan finds faces and stores them locally, with the image they belong to.
Cluster into people
A precision-first grouping pass builds people from faces that are confidently similar.
Merge and name
Merge people sharing a name, then work through unnamed people in the bulk view.
The honest part
Grouping is a suggestion
- Clustering is probabilistic. Some people will be split; some clusters will contain the wrong face.
- The app does not claim identity verification, and it does not know a person from a name alone.
- You can merge and name; the final organisation is yours, not the model’s.
Good questions
Facial grouping, answered.
Does it merge people automatically?
It merges people that share a name where the grouping path supports it, and it offers an Unknown People surface for bulk naming. You remain in control of the names.
Can I correct a wrong group?
Yes. The People page is the correction surface: review, rename and merge as needed.
Does grouping use Windows Live Photo Gallery names?
Where those names are present in the file metadata, the import can use them. It does not invent names for people it cannot match.
Keep exploring
Related features.
Facial recognition
Face detection that stays local.
Find and group the people in your library without sending a face to a server.
Tags
Tags that survive the app.
Read existing keywords, add your own tags, and filter with real counts.
Browse the library
Browse by face, tag and location.
Dedicated people, tags and locations views, plus folder and date filters. Search and object recognition are coming.
See it on your own library
Bring the folder.
Keep the originals.
Download the free beta, point it at a folder and try this on your own photos. No account, no upload, no card.
Windows 11 · Free 5,000-photo beta · No account needed