NNIFT · AR/VR & AIRetail Fashion
Back to roadmap
Session 02 · Unlocked~2 hoursDigital identity & fit technology

Virtual Try-On & the Digital Body

Body scanning, avatars, fit and AI-powered try-on. Last time we learned the vocabulary. Today we make it personal: how your body becomes data, how brands use that data to answer “will this fit me?”, and why this is one of the most exciting — and sensitive — areas in fashion tech.

What you'll get out of today
  • Explain what a 'digital body' is — from tape measure to 3D avatar
  • Tell apart the three families of virtual try-on: AR overlay, 3D simulation and generative AI
  • Understand why 'size L' means nothing, and how fit tech cuts returns
  • Measure yourself properly and build your own avatar
  • Discuss body image, bias and privacy in try-on technology
Today's plan
  1. 0:00Recap of Session 1 & today's goal
  2. 0:10What is a 'digital body'? (tape measure → 3D avatar)
  3. 0:30The three families of virtual try-on
  4. 0:55Why 'size L' means nothing: fit, returns & sizing AI
  5. 1:15Hands-on lab: measure yourself & build an avatar
  6. 1:40Body image, bias & privacy — a group discussion
  7. 1:55Wrap-up, Q&A & homework

A steady ~2-hour flow from concept to hands-on to discussion.

1 · The digital body

From a tape measure to a 3D you

A “digital body” just means your body turned into data a computer can use. That data can be as simple as three numbers (bust, waist, hips) or as rich as a full 3D avatar. Here's the journey almost every try-on tool follows.

Physical bodyDigital body

Measure

Tape measure: bust, waist, hips, inseam

→

Capture

Photos, video or a 3D body scan

→

Reconstruct

AI builds a 3D avatar of you

→

Dress & decide

Drape garments, see fit, get a size

1

Measure

The oldest 'body data' is a tailor's tape measure — bust, waist, hips, inseam. Still the ground truth that everything else tries to predict.

2

Capture

A phone photo (front + side), a short video, or a 3D body scan turns your shape into data — silhouettes and key points a computer can read.

3

Reconstruct

AI converts that capture into a 3D body model (an avatar) with realistic proportions — even estimating measurements you never typed in.

4

Dress & decide

Garments are draped on the avatar so you can see fit and drape, get a size recommendation, and buy with more confidence.

Four ways your body becomes data

Each method trades off accuracy, effort and privacy. There is no single “best” — it depends on what the brand is trying to do.

Tape measure

You physically measure yourself and type numbers into a size guide.

Good: Free, accurate if done right, needs no tech.

Watch out: Easy to measure wrong; boring; no visual of fit.

Two-photo capture

Take a front and side photo in tight clothing; AI predicts your measurements from the silhouettes.

Good: Only needs a phone; surprisingly accurate at scale.

Watch out: Sensitive photos; lighting and clothing affect accuracy.

3D body scan

A booth, depth camera or app rotates around you to build a full 3D mesh of your body.

Good: Very precise; captures true shape, not just numbers.

Watch out: Needs special hardware or careful capture; less accessible.

Statistical avatar

You enter height, weight and gender, and a model predicts a plausible body shape.

Good: Zero friction — no photos, works instantly.

Watch out: It's an average, not you; every body is unique.

2 · The three families of virtual try-on

Not all “try-on” is the same

When a brand says “virtual try-on”, it could mean three very different things. Knowing which one you're looking at tells you what to trust it for.

1

AR overlay

Live camera draws the product onto you. Best for glasses, watches, sneakers & makeup.

2

3D simulation

An avatar + cloth-physics drapes a digital garment to show real fit and tight/loose areas.

3

Generative AI

An AI paints a brand-new photo of you wearing the item — photorealistic, scales to millions.

Roughly ordered simplest → most advanced. Many brands now combine all three.

1

AR overlay (real-time)

How it works
The phone camera watches you live, and computer vision draws the product onto your body or face in real time — glasses on your nose, a watch on your wrist, sneakers on your feet.
Strength
Instant, playful, and 'it's really me'. Great for accessories and makeup that sit on a predictable spot.
Limitation
Struggles with loose clothing and how fabric actually drapes. It's a sticker on top, not a true simulation.

Real example: Warby Parker glasses try-on and Amazon's AR 'Virtual Try-On for Shoes' point your camera at your feet.

2

3D simulation (avatar + physics)

How it works
A 3D avatar is built from your measurements, then a digital garment (a 'digital twin' of the real product) is draped on it using cloth-physics — so you see where it pulls tight or hangs loose.
Strength
Actually answers 'will this fit me?'. A heatmap can show tight vs. loose areas across sizes.
Limitation
Needs accurate measurements and a 3D model of every garment, which is expensive to produce at scale.

Real example: Zalando's virtual fitting room lets you create a 3D avatar from your body measurements and try jeans across sizes.

3

Generative AI (image synthesis)

How it works
An AI image model takes one photo of the garment and one of a person, and generates a brand-new photo of that person wearing it — inventing realistic folds, wrinkles, shadows and drape.
Strength
Photorealistic and works from a single product image, so it scales to millions of items instantly.
Limitation
It's a convincing prediction, not a measurement — it can look perfect yet still be the wrong size for you.

Real example: Google Shopping's virtual try-on uses generative AI (its TryOnDiffusion research) to show apparel on real, diverse models — and now on your own photo.

3 · Why “size L” means nothing

The sizing problem — and how AI helps

There is no global standard for what “M” or “size 48” means. One brand's L is another's M. That single fact drives a huge share of online fashion returns — which are expensive for brands and terrible for the planet.

The problem

Roughly a third of Zalando's returns are size-related — people order the wrong size because guides are inconsistent.

The fix

Predict the right size from your measurements or two photos, then say “order M” before you buy.

The result

Zalando reports ~10% fewer size-related returns with size advice, and up to 40% fewer returns in avatar fitting-room tests.

Big idea: fit tech isn't just a gimmick — cutting returns is one of the clearest ways fashion can save money and reduce waste at the same time.
4 · Watch & learn

See the digital body in action

Three short, real videos — one on the AI research behind try-on, and two on how 3D body avatars are actually built for fashion.

The AI-Powered Online Fitting Room (VOGUE / Google)

What's AI by Louis-François Bouchard

A short, student-friendly walk-through of how Google, MIT & UW used a StyleGAN2-based AI to try clothes on a photo of you (the 'VOGUE' research).

How to create a 3D body avatar

3D Body Tech / IBV Human Analysis Lab

A clear review of body-scanning and avatar methods — where measurements come from and how a 3D body is reconstructed for fashion.

Custom Fit Model Avatars: 3D Body Scanning for Digital Fashion

3DBODY.TECH 2025

Industry talk on real 3D body-scanning workflows used to build fit-model avatars for digital fashion and sizing.

Hands-on lab · Part 1

Try all three families on your own phone

Download a few of these free apps and spend ~10 minutes each. As you go, label what you're using: “Is this an AR overlay, a 3D simulation, or generative AI?”

AR shoe try-onFree

Amazon Shopping

Amazon's 'Virtual Try-On for Shoes' uses AR to show thousands of sneakers on your own feet from every angle.

Try this

Search a supported sneaker, tap 'Virtual Try-On', point the camera at your feet and swap colours live.

iPhone / Android
Generative AI try-onFree

Google (Search app)

Google Shopping's virtual try-on shows apparel on diverse real models — and, in the US, on a full-length photo of you.

Try this

Search a top or dress, look for the 'Try on' badge, and compare how it looks across different body types.

iPhone / Android
AR face try-onFree

Warby Parker

Try eyeglasses on your own face using the front camera and face tracking — the classic accessory AR overlay.

Try this

Open Virtual Try-On, test three very different frames, and notice how well it tracks your face.

iPhone (Face ID)
AR try-on lensesFree

Snapchat

Fashion & beauty brands build AR 'lenses' — sneakers, hats, makeup, bags — that track your face and body live.

Try this

Search a brand's shopping lens (e.g. shoes or makeup) and see real-time body/face AR overlay.

iPhone / Android
Hands-on lab · Part 2

Measure yourself properly & build your body profile

Before any avatar can represent you, someone has to measure a body. Do it once, do it right, and you'll use these numbers again in later sessions.

1 · Measure

With a soft tape, record height, bust/chest, natural waist, hips and inseam. Keep the tape snug, not tight, and stand relaxed.

2 · Compare

Pick one brand you own and look up its size chart. Which 'size' do your numbers actually map to? Note any surprises.

3 · Go digital

Use a two-photo size tool or a virtual fitting room and compare its guess to your tape numbers. How close was it?

Tip: measure in front of a mirror, or ask a friend. The single biggest source of bad online fit is people measuring themselves incorrectly.
Hands-on lab · Part 3

Create your own 3D model — the exact apps & websites

Now make a digital body of your own. These are the specific, beginner-friendly tools we'll use — pick one and bring your result to class. Start with Ready Player Me if you've never done this before.

Web · any browserFree for personal use

Ready Player Me

The easiest starting point: take a selfie (or pick a preset) and get a full-body, game-ready 3D avatar you can customise with clothes, hair and accessories.

Try this

Go to readyplayer.me, upload a selfie, style your avatar, and download the 3D model (GLB).

Web · any browserFree to create & export

Avaturn

Uses generative AI to turn a single photo into a recognisable, realistic 3D avatar of you — with an adjustable body shape — that exports to Blender, Unity or Unreal.

Try this

Visit avaturn.me, upload a front + side photo, tweak the body, and export your avatar.

iOS appFree tier (paid upgrades)

in3D: Avatar Creator Pro

Scans a short video selfie into a rigged 3D avatar of your face and body in about a minute — then lets you try on clothes and export the model.

Try this

Install the app, record the guided video selfie, and see yourself as a 3D avatar.

iPhone / AndroidFree to try

Polycam / Scaniverse

Photogrammetry scanners (from Session 1). Instead of an avatar, use these to 3D-scan a real garment, shoe or accessory into a model you can dress an avatar in.

Try this

Scan a shoe or jacket by circling it slowly, then rotate the finished 3D object.

Privacy note: these tools use a photo or video of you. Read what each one does with your image, and never upload photos of anyone who hasn't agreed to it.
Discuss · Body image, bias & privacy

The digital body raises real questions

This tech is powerful, but it touches people's bodies, self-image and personal data. Discuss these in small groups — there are no “correct” answers.

  • 1If an AI try-on always makes clothes look flattering, is it helping you decide — or just selling harder?
  • 2Google says it deliberately used models across sizes XXS–4XL and many skin tones. Why does the choice of training data matter for who a tool works well for?
  • 3Two-photo and 'be your own model' features ask for near-nude photos of your body. Who should be allowed to store them, and for how long?
  • 4Could seeing a hyper-idealised avatar of yourself affect body image — positively or negatively?
Study material · Glossary

Keep these terms handy

The vocabulary of the digital body. Skim now, and come back whenever a word feels fuzzy.

Digital Body

Your body as data

Any digital representation of your body — from a set of measurements to a full 3D avatar — that software can use to show fit and drape.

Avatar

Your 3D stand-in

A 3D model of a person used to try garments on virtually. It can be built from your measurements, a scan, or a statistical estimate.

3D Body Scan

A 3D mesh of you

Capturing the full shape of your body as a 3D mesh using depth cameras, a scanning booth, or careful phone capture.

Anthropometry

The science of body measurement

The systematic measurement of the human body — the data that sizing systems and avatars are ultimately built on.

Digital Twin (garment)

A 3D copy of a product

A precise 3D model of a real garment that can be draped on an avatar with cloth-physics to simulate fit.

Cloth Simulation

Physics for fabric

Software that calculates how fabric stretches, folds and drapes on a body, so a 3D try-on shows realistic fit — not a flat sticker.

Diffusion Model

AI that paints pixels

A generative-AI technique that builds an image pixel-by-pixel. It powers photorealistic try-on like Google's TryOnDiffusion.

Size Recommendation

'Order size M'

An AI prediction of your best size for a specific item, based on your measurements, purchase history and the garment's data.

Fit Heatmap

Tight vs. loose, visualised

A colour overlay on an avatar showing where a garment sits tight or loose, helping you choose between sizes.

Monk Skin Tone Scale

A 10-shade inclusivity scale

A skin-tone scale Google uses to select diverse models, so a try-on tool represents a wide range of people fairly.

Sources & further reading

Where these facts come from

Every brand claim on this page links to a primary source — company press rooms and peer-reviewed research. Great for your own projects.

Before next session

A little homework

  • 1Try ONE try-on tool from each of the three families (AR overlay, 3D simulation, generative AI). Save a screenshot of each.
  • 2Measure yourself with a tape and compare your numbers to one brand's size chart. Write down which size you 'should' be vs. what you usually buy.
  • 3Write 4 sentences: which family felt most useful, which felt most 'off', and one privacy concern you'd want a brand to address.

Previous

Session 1 · Seeing the Future of Fashion

Next up

Session 3 · AI as a Design Collaborator