Try it onNot your size? Doesn't matterYour closet, digitisedFewer returnsTry it onNot your size? Doesn't matterYour closet, digitisedFewer returns
The problem
Shopping for clothes is still a guess.
You cannot see how it sits on your body at the exact
moment you have to decide. So you order three sizes, or you queue outside a changing room. Both are terrible.
01
Buying blind
A flat photo on a stranger's body tells you nothing about yours. Order three, return two.
02
The queue tax
Wait, change, look, repeat. One garment at a time. Everyone gives up by the third.
03
Nothing to wear
A full cupboard and the same panic every morning. Outfits you own, never found.
04
Tried by twenty
Staff refold rejects all day — and you buy something twenty strangers already wore.
25–35%
Return rate on fashion e-commerce in India. Highest-friction category online.
60%
Of India's fashion is still bought offline, where the fitting room is the bottleneck.
$21.6B
India fashion e-commerce in 2025, growing ~24% a year. Friction scales with it.
Sources: IBEF · Nexdigm India Fashion Retail Outlook 2030 · CoherentMI India Fashion Ecommerce 2025–2032.
How it works
Upload two things. That's the whole app.
Not a diagram — the actual flow. Pick a photo, pick
a garment, hit generate.
Your photo
STEP 01
A photo of you
Any phone, decent light, arms relaxed. The same selfie you'd send a friend works fine.
The garment
STEP 02
A photo of the fit
Screenshot it off any store, or shoot the thing lying on your bed. Both work.
TyF outputHit generate and it appears hereReading the pose
STEP 03
You, wearing it
Around ten seconds later. Your face, your hair, your background — completely untouched.
Eight sample pairs, each rendered in advance by the same engine — nothing here is a mock-up.
Uploading your own photos opens with early access.
Real output
Every image below came out of the engine.
No retouching, no cherry-picked studio lighting.
Ordinary photos in, these out — and the two photos that went in are sitting in the corner of each one.
Our working proof of concept, recorded end to end. Every model
in the pipeline was trained in-house — this is not a wrapper around someone else's API.
0.795
SSIM · structural match
0.217
LPIPS · perceptual match
44.06
FID · realism
The product
One engine. Two ways in.
The app that already knows your wardrobe
Shoot what you own once. After that every "what do I wear" gets answered in
seconds — from clothes already in your cupboard, or anything you're about to buy.
Try before you buy. Screenshot from any store, see it on yourself.
Digitise your closet. Photograph each piece once, mix forever.
Occasion mode. Wedding, interview, first date — outfits from what you own.
Share the fit. Send it to the group chat before you commit.
Drop the widget onto your listings, or call the API from your own stack. For
stores, kiosk mode turns a tablet at the rack into a fitting room with no queue.
Product-page widget. A script tag and a catalogue feed.
REST API. Two image URLs in, a result out, usually under ten seconds.
All prices in INR, exclusive of GST. Business tiers quoted on catalogue size and monthly volume.
Built in BengaluruOwn model, own engineNo body scanNo measurementsBuilt in BengaluruOwn model, own engineNo body scanNo measurements
The team
Three founders.
TyF is built by Triye Technologies, a Bengaluru studio
working on applied computer vision. We built the engine before raising on it.
Akhilesh N Naidu
Co-founder · AI & Product
Owns the engine — architecture, training runs, evaluation, and everything that turns two uploads into a result.
Akshay R Kumar
Co-founder · Business
Owns the market — product vision, customer discovery, enterprise sales, partnerships and fundraising.
Aditya Belludi
Co-founder · Deployment & Ops
Owns the delivery — infrastructure, deployment, customer relations and compliance.
Questions
The obvious ones.
No. One ordinary front-facing photo is the entire input. No depth sensor, no turntable,
no measurement form.
On 200 held-out test pairs: SSIM 0.795, LPIPS 0.217, FID 44.06. In plain terms — body
geometry and garment placement come out right, colour and print transfer faithfully, and there's a measurable
gap to studio photography in the finest texture detail. We publish numbers, not adjectives.
Strongest on upper-body garments with a front-facing subject and relaxed arms — tees,
shirts, knits, overshirts. Full-body framing, dresses and heavily draped Indian wear are the next milestone,
limited by training-data coverage rather than the architecture.
They generate your result and build your own wardrobe library. They are not sold, and not
used to train models without explicit opt-in. Delete any image or your whole account at any time — that removes
the sources and the generated results together.
Yes — that's the standard business setup. Point us at a catalogue feed and the widget
renders try-on on your product pages, or call the REST API from your own stack.
The proof of concept works now — the demo above is real output. Early access opens in
waves to waitlist members, consumers first, with retail pilots in parallel.
Early access
Stop guessing what fits.
Founding members get locked pricing,
first access to the closet engine, and a direct line to the people building it.
✦ You're in.We'll write the moment access opens.
No spam. One email when it's ready. Running a store or a brand?
Real output
Two photos in, one image out
Exactly what went into the engine, and exactly what came back. Nothing was retouched
between them.
Your photo+The garment=TyF output
For business
Book a pilot
Tell us what you sell and roughly how much of it. We'll come back within two working days with a
scope and a price.
