The Technology

Two photos become a fit answer in seconds

Thinx3D rebuilds a shopper's body shape from front and side phone photos, then compares it with graded garment measurements to forecast the item's fit.

The Process

Three steps to a confident size

In 60 seconds, shoppers get fit guidance and you get returns driven by fit, not guesswork.

1
Shopper takes two photos

Front and side, standing naturally. No special hardware, lighting, app, or equipment needed, only the phone in their pocket.

2
Thinx3D builds a 3D body model

Photogrammetry reads silhouette, shadow, and pose depth cues to form a parametric 3D model calibrated against real body dimensions.

3
The model recommends a size

We compare each measurement with your graded specs across sizes, then show the best match on the product page with a clear explanation.

chest waist hip 3D body model thinx3d.com
Under the Hood

Inside the fit engine

Every Thinx3D pipeline phase targets accuracy at the point where fit judgments typically fail.

01

Photos become clean body silhouettes

A lightweight widget on your product page accepts two shopper photos. Our pipeline removes background noise with a segmentation model trained across diverse body types, producing clean silhouettes for modeling.

front photo side photo
02

Parametric body model reconstruction

Contours from both views enter constrained optimization, fitting a parametric 3D body model to the observed outlines. It delivers circumference and length measurements accurate within 1.8 cm across key fit zones, validated against a corpus of 3D body scans.

94.2cm 76.8cm 99.1cm
03

Comparing specs and recommending a size

We test the shopper's measurements against your graded garment specs at every available size. A score balancing relaxed or fitted preferences with ease allowances selects the best fit, then displays it with a zone-by-zone breakdown.

Fit analysis for Size M
Chest
Good fit
Waist
Relaxed
Hip
Snug
Recommendation Size M, or Size L for a looser hip
Accuracy

Measurements built for real bodies

Thinx3D was tested against 600 participants spanning body type, ethnicity, age, and clothing condition. These figures come from validation, not selected demos.

1.8 cm
Mean absolute error for chest, waist, and hip measurements
94%
Accuracy of size recommendations versus shopper-confirmed best fit
600+
Body scan corpus, diverse body types and demographics
Returns impact

Fewer size mistakes, fewer returns

Guessing a size creates a consistent return pattern. Shoppers using Thinx3D guidance return at about half the rate of shoppers choosing without fit help. The dashboard shows a representative early-access brand partner before-and-after.

The gains build: fewer returns cut reverse-logistics expense, lift net revenue per order, and make repeat purchases more likely among first-time buyers.

Return rate by cause -41% total returns
Before
72%
After
31%
Category
45%

Ready to apply the engine to your catalog?

Thinx3D connects Shopify and custom platforms through a lightweight SDK. Integrations launch within a week. Early access is limited to select brands each quarter.