Photograph your group, let the page find the faces, name each friend — and the names come back printed on cans, ordered so the middle one reads closest and the outer ones step away.
The mechanic only works if the composite looks like it belongs in the photo. Any group, any lighting, up to five people, on a phone camera: the faces have to be found, the names have to attach to the right ones, and the cans have to sit in the picture at a believable depth rather than in a row across it.
Detection places the cans; the person holding the phone names them. The machine does the tedious part and the human does the part it would get wrong.
Scale and placement come from each can’s distance from the centre of the group, so the middle one is largest and the ones either side taper — and the same curve holds whether there are two names or five, which is what keeps it from looking authored case by case.
The photo, the cans and the group’s title are drawn together into a single picture, because the thing people send each other is the point of the campaign.
Title the photo, save the offer, share it — the reward is attached to the artefact rather than parked on a screen of its own.
A photo mechanic that holds up on a real group photo taken in a restaurant, and produces something people had a reason to send on.