← Back to portfolio

AI Photobooth

A generative-AI portrait kiosk for live corporate events in Singapore: QR check-in, AI-generated branded portraits with outfit and car selection, automatic retry and download by QR, about 80–100 guests a day.

AI photobooth kiosk at a shopping-mall event, guest scanning the QR code to download the portrait

Overview

Project details

On a 9:16 kiosk screen a guest accepts the terms, confirms a phone number with an SMS code, picks a look, is photographed on a countdown and takes the result home by scanning a QR code.

Only the face is replaced; the scene and outfit of the approved template stay untouched. A face swap on Replicate builds the portrait, WAN 2.2 animates the still, and ffmpeg prepares an MP4, a WebM boomerang and a looping GIF. A photo is ready in about 12 seconds, an animation in about 27.

Generation runs in its own process, so reloading the kiosk tab does not drop a job; failed generations retry automatically. The catalogue of looks, takeaway branding, opening hours, limits and retention live in an operator panel.

  • About 80–100 guests a day
  • Photo in ~12 s, animation in ~27 s
  • Automatic retry and QR download