Hennessy – Made For More + Glitz
AI-powered runway video experience for Hennessy's luxury brand campaign
Role
Full Stack Developer & AI Engineer
Year
2024
Tech Stack
9 Technologies
Status
Completed
Overview
Hennessy's 'Made For More + Glitz' campaign needed a digital experience that put consumers inside the brand world. I built an AI-powered runway video booth where users upload a selfie, pick a designer look and a runway scene, then walk away with an 8-second photorealistic video of themselves on the runway — 9:16, shareable, generated in under 3 minutes. A 6-stage AI pipeline (Gemini 2.5 + Vertex AI Imagen + Grok Imagine) runs from face validation to video animation, with real-time SSE progress streaming and full brand IP protection via prompt sanitization.
The Problem
Hennessy needed an immersive digital experience for their 'Made For More + Glitz' luxury campaign — something that put consumers inside the brand world, not just watching it. Traditional photo booths and static activations couldn't create the shareable, personalised moments the campaign demanded.
Screenshots
Demo
Challenges
Preserving exact facial identity across a 3-stage AI generation pipeline (composite image → runway scene → video animation) without drift or distortion.
Building a pipeline that takes up to 3 minutes end-to-end without losing the user — long-running AI work that had to feel live, not frozen.
Handling AI content moderation blocks mid-pipeline without failing the whole request and sending the user an error screen.
Keeping Hennessy brand IP out of AI-generated content per legal requirements — the experience is Hennessy-branded, but the videos must be clean of trademarked terms.
Solutions
Used Gemini's inline reference image support to anchor facial identity at every generation stage — the user's selfie is passed as a reference input alongside each AI prompt.
Implemented Server-Sent Events (SSE) for real-time pipeline progress streaming — the user sees each stage complete live, not a blank loading screen.
Added retry logic with fresh composites on moderation block (MAX_CHAIN_ATTEMPTS = 2) — a new composite often bypasses the restriction on the second pass.
Aggressive prompt sanitization strips Hennessy trademarks, cognac references, and founding year from all AI prompts before they reach any generation model.
Key Highlights
6-Stage AI Pipeline: Validation → composite image (Gemini + Imagen) → runway scene placement → video prompt generation → animation (Grok Imagine) → delivery.
45 Unique Combinations: 9 curated fashion looks × 5 runway scenes, all styled against Hennessy's official brand color system (Limestone, Hennessy Red, Gold).
Real-Time SSE Streaming: Users watch each pipeline stage complete live — no spinner, no silence, no abandoned sessions.
Brand IP Protection: Every AI prompt is sanitized of trademarks and brand terms before generation, satisfying Hennessy's legal requirements.
Scale & Scope
9
Fashion Looks
5
Runway Scenes
8s
Video Length
6
AI Pipeline Stages
Technology Stack
Outcome
Delivered a live branded experience for Hennessy's 'Made For More + Glitz' campaign, letting consumers step into a luxury runway moment through a multi-stage generative AI pipeline — photorealistic, shareable, and IP-compliant.
