How browser-based AR foot scanning works
A guided 12-step scan turns a phone camera into a foot-measuring tool. It captures foot length and ball girth to within ±2 mm of a podiatrist scan, then recommends the right size for a specific shoe — all in the browser, with no app to install.
Last updated June 2026 · ARFits Editorial
In short
The shopper opens a scan from the product page. On iOS it launches as an App Clip from the web; on Android it runs through WebXR. A 12-step guided capture takes about 30 seconds and measures foot length and ball girth to within ±2 mm of a podiatrist scan (≈99% accuracy). The system then shows a per-model size recommendation and a fit heatmap, and measurements stay on the device by default.
The 12-step guided scan
The scan is "guided" because the interface walks the shopper through a fixed set of capture positions. Following the guidance is what keeps the result repeatable and accurate — the steps below describe the flow from open to recommendation.
-
1
Open from the web
Tap a scan link or QR on the product page. iOS launches an App Clip; Android opens a WebXR session. Nothing is installed from an app store.
-
2
Grant camera access
The browser requests the rear camera. Capture and processing run locally on the device.
-
3
Set up the environment
Stand on a flat, contrasting floor in good light. Even lighting and a clean background help the system separate the foot from the floor.
-
4
Frame the foot
Position the foot inside the on-screen guide. The interface confirms when the foot is correctly in frame before capture begins.
-
5–10
Guided multi-angle capture
The shopper moves the phone around the foot through a sequence of prompted positions. Capturing the foot from multiple angles is what allows length and the wider ball-girth dimension to be reconstructed, rather than guessed from a single photo.
-
11
Compute measurements
The captured frames are combined into foot length and ball girth, measured to within ±2 mm of a podiatrist scan.
-
12
Recommend a size
A recommendation card maps the measurements to the selected model's size, with a fit heatmap and one-tap half-size adjustment.
The whole flow takes about 30 seconds. Steps 5–10 represent the guided capture positions; the exact prompts vary slightly by device.
What the scan measures
Foot length
The heel-to-toe length is the primary input for size. It maps most directly to EU/US/UK sizing on a per-model basis.
Ball girth
The circumference around the widest part of the foot (the ball). Two feet of the same length can need different sizes if one is wider — ball girth is what catches that.
Capturing both length and ball girth is why a guided AR scan can recommend a size with more confidence than a length-only template or a self-reported size chart.
How ±2 mm accuracy is achieved on a phone
Sub-3 mm accuracy from a consumer phone comes from combining several things rather than a single trick:
- •Guided multi-angle capture. Multiple prompted viewpoints give the system enough information to reconstruct 3D foot dimensions instead of estimating from one photo.
- •On-device sensing. The scan uses the phone's camera and motion sensing to establish real-world scale, so a pixel measurement maps back to millimetres.
- •Step-by-step guidance. Because the interface enforces the capture sequence, the inputs are consistent from shopper to shopper — the main reason results are repeatable.
- •Validation against a reference. The ±2 mm figure is stated relative to a podiatrist scan, which is the practical "ground truth" for foot measurement.
Accuracy depends on following the on-screen guidance, adequate lighting, and a suitable surface. Reported accuracy is about 99% versus a podiatrist scan; it is a measurement aid, not a medical device.
App Clip (iOS) vs. WebXR (Android)
| Aspect | iOS — App Clip from web | Android — WebXR |
|---|---|---|
| How it launches | An App Clip is invoked from the web — a lightweight, on-demand experience. | Runs directly in the mobile browser via the WebXR API. |
| App-store install | Not required. | Not required. |
| Entry point | Scan link or QR on the product page. | Scan link or QR on the product page. |
| Where data lives | On the device by default. | On the device by default. |
The two paths exist because the platforms expose camera-based AR differently. For the shopper, both feel the same: tap, scan, get a size — no download.
The fit heatmap and recommendation card
Measurements alone do not tell a shopper what to buy. The recommendation step turns them into a decision for a specific model.
Recommendation card
Suggests the best size for the selected shoe, based on the scan plus the model's last and fit profile.
Fit heatmap
Teal shows where the shoe fits comfortably; coral shows where it runs tight. The shopper sees why a size is suggested.
One-tap half-size
Tapping a half-size up or down recomputes the heatmap instantly, so shoppers can weigh a snug vs. roomy preference.
Privacy: measurements stay on the device
By default, the scan's measurements stay on the shopper's device. Because the experience runs in the browser through an App Clip or WebXR — rather than a persistent installed app collecting data in the background — the foot data does not need to leave the phone to produce a size recommendation. This is a meaningful difference from sending photos of your feet to a server.
A secondary benefit: visual AR try-on
The same camera flow can also place the shoe on the foot in augmented reality. This is a secondary feature to the sizing job, but a useful one: it lets shoppers judge look and fit together — does this sneaker suit me, and will this size sit right — before they buy. Sizing remains the core value; the visual try-on is the confidence layer on top.