A friend who's a corporate photographer came to me with an interesting problem: aligning multiple portrait photos for team pages, yearbooks, directories, so that faces are consistently positioned.

Fundamentally, when you take dozens of photos in a day, they are likely to be all slightly different in terms of crop.

As I had done some work with facial feature recognition, my friend asked if the same tech could be used for his problem... and here's why FaceCrop came to be :-)

How does it work:

- drop in your photos

- FaceCrop detects faces using face-api.js or tracking.js

- FaceCrop suggests crops

- fine-tune the crops individually or globally

- export the crops in a .zip file.

To note, everything runs client-side on the browser: no uploads, no server-side, no accounts. It was mostly developed using some code I had from a previous project and some Claude Code.

Would welcome constructive feedback most importantly on the crop adjustment UI/UX, but also on the detection accuracy, suggested crop shape, etc.

http://facecrop.puntofisso.net/