Home Works Info
← Back to All Work

Tom Ford Ombré Leather — Luxury in a Face Filter

A Spark AR Instagram filter for Tom Ford Beauty — editorial-grade face tracking and segmentation serving a live event photo booth and the brand's global audience.

Role
AR Designer & Spark AR Developer
Visual concept · Spark AR development · Face tracking · Person segmentation
Timeline
3 months
Platform
Instagram (Spark AR)
Team
Vincent MeiCreative Director
Product by
Collaborators
Ombré Leather AR filter
The Problem

Luxury beauty standards, inside real-time face tracking.

Business

  • One filter serving an event booth and Instagram
  • Editorial quality worthy of the brand

User

  • Flattering on diverse skin tones and face shapes
  • Zero lag on mid-range phones

Technical

  • Strict Spark AR asset budgets
  • Person segmentation varies by device
Why This Was Hard
01

The luxury gap

What the brand expects visually versus what real-time tracking can deliver.

02

Aggressive optimization

Every texture squeezed under Spark AR's asset ceiling.

03

Diverse faces

One mask that sits naturally on every skin tone and face shape.

04

Dual deployment

The same build powering a photo booth and a social filter.

My Role

I designed and built the filter — concept through Spark AR deployment.

Owned

  • Visual Concept
  • Spark AR Development
  • Face Tracking & Segmentation
  • Event Deployment

Collaborated with

  • Creative Director
  • Tom Ford Beauty (client)
Key Decisions

Translating a fragrance into light.

01
Decision 01 — Concept

Three directions, one hero

Challenge

Ombré Leather is tactile — shadow, texture, smoke — not screen-native.

Decision

Pitched Smoke Veil, Leather Mask, and Golden Hour; shipped Leather Mask with Golden Hour's glow.

Reasoning

Structure carries the brand; warm radiance carries the fragrance's amber character.

Outcome

A filter that reads as campaign imagery, not a gimmick.

Leather Mask with golden-hour glow
Tom Ford Ombré Leather — decision 1 detail
02
Filter in use
Tom Ford Ombré Leather — decision 2 side
Decision 02 — Engineering

Optimize for the worst phone in the room

Challenge

Luxury cannot lag — and event guests carry mid-range devices.

Decision

Texture budgets, segmentation fallbacks, and performance floors set before visual polish.

Reasoning

A dropped frame breaks the illusion faster than a smaller texture ever will.

Outcome

Smooth tracking across iOS and Android at the live event.

03
Decision 03 — Deployment

One build, two stages

Challenge

An event photo booth and Instagram distribution have different lighting, framing, and attention spans.

Decision

A single filter tuned for both contexts rather than two divergent builds.

Reasoning

Press moments and social reach had to show the identical brand image.

Outcome

Coverage in Vogue, WWD, and Flaunt from one asset.

Tom Ford Ombré Leather — decision 3
Tom Ford Ombré Leather — decision 3 detail
Innovation

Constraint matrix

Every brand requirement paired against Spark AR feasibility before creative lock.

Editorial segmentation

Users lifted from their background into a designed theme environment.

Booth-to-feed pipeline

The same filter serving in-person guests and global audiences.

Impact
Business
Press
Featured in Vogue, WWD, and Flaunt
Product
2-in-1
One build serving the live booth and Instagram
Product
Smooth
Performance floor held on mid-range devices
Team
Repeatable
Constraint-matrix method reused on later brand AR work
Reflection

The trade-off — I locked performance budgets before visual exploration. It constrained the concepting space, but luxury AR that stutters is worse than none at all.

What I'd improve — the mask's fit at extreme face angles. More tracking-edge testing would have tightened the last five percent.

Next time — for brand AR, present concept directions against a feasibility matrix — clients choose confidently when beauty and constraints are shown together.

Project Showcase
Tom Ford Ombré Leather — showcase a
Tom Ford Ombré Leather — showcase b