JI.
AI Photobooth - AI-generated portrait experience
Back to Work

AI Photobooth

AI / Experiential / Interactive2026 - Present

A full-stack AI photobooth for pop-ups, museums, and brand events: tap an iPad and walk away with a portrait that keeps your likeness in a new world, no app download and no training on user data.

~15sGeneration Time
170+Cards, One Event
0App Downloads
0Data Used for Training
RoleCreator / Lead Developer
PlatformWeb App (Next.js)
TypeExperiential Installation
HardwareiPad on Stand
AI EngineLatest Image Gen APIs
IdentityFine-Tuned Likeness Lock
PrivacyZero Training on User Data

The Experience

01

Walk Up

An iPad on a stand runs the web app, no sign-up or download.

02

Capture

Tap, 3-second countdown, flash, photo captured.

03

Preview

The guest confirms the capture and sends it to generation.

04

Generate

The AI transforms the photo in ~15 seconds while loading text explains what is happening.

05

Take Home

The finished image lands on a branded page with a QR code to download, share, or revisit.

How It Works

Identity Preservation

The AI is fine-tuned to keep the guest's likeness - facial features, skin tone, expression - in new environments, styles, and scenarios. Single pass, no pre-training, and no user data stored or used for training.

Infinite Scenarios

Courtside at the Super Bowl, inside a cartoon, on stage at a concert, next to a celebrity. Each preset is a prompt configuration for scene, style, and compositing rules, so new scenarios spin up for any brand, event, or theme.

Owned Ecosystem

Capture app, generation pipeline, display, QR landing page, and distribution are all built in-house. That means full control of branding, privacy, and monetization, with no photobooth vendor in the loop.

Built for Activations

Made for pop-ups, museum installations, brand activations, festivals, and retail. The iPad-on-stand setup is portable and only needs WiFi.

Generated Outputs

Super Bowl sideline AI transformation with Bad Bunny
Squid Game AI transformation - placed into the Red Light Green Light scene
Cartoon style AI transformation - SpongeBob Krusty Krab
Mr. Pink selfie - Flatiron Building AI photobooth activation
Concert crowd surf AI transformation
Super Bowl sideline AI transformation with Bad Bunny

Proven in the Field

4th Annual Friends & FANmily Game

An outdoor, all-ages back-to-school community day at Victory Field in Queens, August 2026, put on by Mets creators and sponsors including the Fanatics Fan Advisory Network and AmazinVntg. I ran the trading-card booth: one selfie in, a card of yourself in a batting pose out, sponsor logos on the back.

170+ cards in one day, zero timeouts, no laptop on site; copy tweaks and a faster countdown went live from a phone mid-game. amNY covered the day; the “virtual trading card activation” in their write-up is this booth. Read the amNY story.

Designed for the Bad Day

Image APIs fail a few percent of the time, so every failure path gets an instant retry and plain on-screen copy. Stress tests set the timeout budget; the first real guests set the countdown speed and changed the capture prompt from batting pose to selfie, since the system poses the guest.

The input: one selfie at the booth
The input: one selfie at the booth
The output: a 1987-style card, posed by the system
The output: a 1987-style card, posed by the system
Every guest centered, uniformed, named
Every guest centered, uniformed, named

Cards above come from the event pipeline; the people in them are generated, so no guest's face appears without asking.

170+Cards in one day
0Timeouts
0Dead ends for guests
100%Operated remotely

Use Cases

Brand Pop-Ups

Product launches, store openings, sponsored events, with branded presets and a shareable takeaway.

Museums & Exhibitions

Visitors become part of the exhibit: inside famous paintings, historical moments, or imagined futures.

Festivals & Live Events

High-throughput for concerts, food festivals, and cultural events; the QR page keeps working after.

Tech Stack

Next.jsReactWebRTCAI Image Generation APIsTailwind CSSFramer MotionVercelServer ActionsCanvas APIQR Code Generation

More Projects