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.
- 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
- 0:00Recap of Session 1 & today's goal
- 0:10What is a 'digital body'? (tape measure → 3D avatar)
- 0:30The three families of virtual try-on
- 0:55Why 'size L' means nothing: fit, returns & sizing AI
- 1:15Hands-on lab: measure yourself & build an avatar
- 1:40Body image, bias & privacy — a group discussion
- 1:55Wrap-up, Q&A & homework
A steady ~2-hour flow from concept to hands-on to discussion.
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.
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
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.
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.
Reconstruct
AI converts that capture into a 3D body model (an avatar) with realistic proportions — even estimating measurements you never typed in.
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.
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.
AR overlay
Live camera draws the product onto you. Best for glasses, watches, sneakers & makeup.
3D simulation
An avatar + cloth-physics drapes a digital garment to show real fit and tight/loose areas.
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.
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.
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.
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.
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.
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.
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?”
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.
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.
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.
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.
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?
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.
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).
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.
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.
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.
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?
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 dataAny 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-inA 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 youCapturing 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 measurementThe 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 productA precise 3D model of a real garment that can be draped on an avatar with cloth-physics to simulate fit.
Cloth Simulation
Physics for fabricSoftware 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 pixelsA 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, visualisedA 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 scaleA skin-tone scale Google uses to select diverse models, so a try-on tool represents a wide range of people fairly.
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.
Amazon — Virtual Try-On for Shoes
Amazon Press Center (Jun 9, 2022)
Official announcement of Amazon Fashion's AR shoe try-on for iOS, covering thousands of styles.
Google — AI virtual try-on for apparel
The Keyword, Google Blog (2023)
Google's generative-AI try-on shows clothes on real models across sizes XXS–4XL using the Monk Skin Tone Scale.
Google — 'Try it on' with your own photo
The Keyword, Google Blog (I/O 2025)
Update letting shoppers upload a full-length photo to try billions of apparel items on themselves.
TryOnDiffusion — research paper
Google Research (CVPR 2023)
The diffusion model ('a tale of two U-Nets') behind Google's generative virtual try-on.
Zalando — size advice from body measurements
Zalando Corporate (Jul 17, 2023)
Two phone photos predict a customer's measurements to recommend the right size; cut size-related returns ~10%.
Zalando — 3D avatar virtual fitting room
Zalando Corporate (Oct 17, 2024)
Avatars built from real body measurements; reported up to 40% reduction in return rates in testing.
Walmart — 'Be Your Own Model'
Walmart Corporate (Sep 15, 2022)
Upload your photo to see 270,000+ items on your own body, powered by the Zeekit acquisition.
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