THENN

The Challenge

In AI photography the real difficulty is likeness: however beautiful the frame, it is worthless if it does not look like the person. On top of that came multiple people in one frame keeping their own faces, eight languages, store review rules and a transparent consent flow around face data.

The Approach

1. // 01An identity keeping generation pipeline preserves real facial features at every step, matching each face to its own reference in multi-person scenes.
2. // 02More than a hundred styles were crafted by hand: Santorini sunsets, Cappadocia balloons, Paris nights, wheat field maternity, knit bowl newborn, golden hour family portraits.
3. // 03Two quality modes were added, same model and same likeness, one for the finest detail and one for speed.
4. // 04Credit based packs replaced subscriptions: one credit, one photo, bought once.
5. // 05An explicit face data consent flow was written, showing what is sent, who processes it and how long it is kept, before the first upload.
6. // 06Release in eight languages, store review and version management were handled end to end.