DYNAFIT

.Personal PROJECT

Democratizing 4D Anthropometry for Ultra-Personalization

Ultra-Personalisation

Category:

Ultra-Personalisation

Personal Project

Read:

11 mins

Tag:

AI , Computer Vision, Anthropometry, Dynamic, Personalisation

Completion Date:

Still Ongoing

GOAL

To architect a system that captures the body’s changing biomechanics and surface deformation using accessible smartphone video, democratizing high-fidelity 4D anthropometry.

.THE PROBLEM

We Are Designing for Statues

In the fields of Medical Design and Performance Wearables, there is a critical "validity gap." We currently use static measurements (calipers and tape measures) to design products for a body that never stops moving.

The Accessibility Barrier

True 4D scanners (capturing 3D shape over time) are expensive, lab-grade machines, limiting data collection to specialized facilities. This makes individual-specific personalization inaccessible for most patients and consumers.

The Dynamic Reality

In reality, the human shape deforms, the dimensions of the body change over time. For Example, Human back expands by over 12cm during flexion. A forearm swells significantly when gripping a tool

The issue with static scans

Static scans fails to accommodate the dynamic changes in a individual. We use generalized parameters for the dynamic variations. Making designs less personalized.

The Accessibility Barrier

The issue with static scans

Static scans fails to accommodate the dynamic changes in a individual. We use generalized parameters for the dynamic variations. Making designs less personalized.

The Dynamic Reality

In reality, the human shape deforms, the dimensions of the body change over time. For Example, Human back expands by over 12cm during flexion. A forearm swells significantly when gripping a tool

The Accessibility Barrier

The issue with static scans

Static scans fails to accommodate the dynamic changes in a individual. We use generalized parameters for the dynamic variations. Making designs less personalized.

The Dynamic Reality

In reality, the human shape deforms, the dimensions of the body change over time. For Example, Human back expands by over 12cm during flexion. A forearm swells significantly when gripping a tool

The Accessibility Barrier

The issue with static scans

Static scans fails to accommodate the dynamic changes in a individual. We use generalized parameters for the dynamic variations. Making designs less personalized.

The Dynamic Reality

In reality, the human shape deforms, the dimensions of the body change over time. For Example, Human back expands by over 12cm during flexion. A forearm swells significantly when gripping a tool

Mannequin generated using Dined.nl

.Solution

The Solution

After researching about the different frameworks available and designing and testing the different existing solutions in computer Vision and biomechanics , I engineered Pipeline 2.0 (This is the 4th iteration of the pipeline). This architecture decouples the Visual reconstruction from the Biomechanical validity and introduces a "Human-in-the-Loop" workflow using SAM 3D from META.

DATA ACQUISITION & CALIBRATION

Visual Reconstruction

Biomechanical Registration

Physics-Based Contact

DYNAMIC Metric Extraction

Smartphone Video

Focus Lock

ChArUco Calibration

Undistorted frames

SAM 3D Body

Consistent Mesh

Occlusion Handling

MHR Topology

SKEL Integration

Inverse Kinematics

Anatomical Correction

SUPR Integration

Object and realistic ground interaction

Custom Metric Analysis Over Time

The pipeline extracts the mesh from a video or an image and reconstructs it using SAM 3D, SKEL and SUPR Frameworks. At the end you end up with a Consistent 4D Mesh Sequence, Biomechanical Joint Data, Contact & Pressure Maps, Dynamic Metric Reports

.Use CASES

WHAT IT ENABLES.

Prosthetics & residual limb design
4D anthropometry maps limb deformation through gait cycles, enabling sockets that adapt to daily volume change and pressure hotspots.
Occupational PPE & firefighting suits
4D scans reveal how protective suits shift, compress, and form air gaps during real tasks, enabling safer, less restrictive PPE
Respiratory masks & pilot gear
4D face meshes capture jaw, cheek, and breathing motion, guiding mask seals that stay tight during speech and strain
Specialized athletic footwear & 4D feet
4D foot scans capture full gait deformation, informing soles, toe boxes, and counters tailored to each runner’s pressure paths.
Ultra-personalized athletic wear
4D body data tracks circumference and shape change in motion, enabling sport-specific compression zones tuned to each athlete.
Pregnancy & maternity wear
4D body sequences map trimester-specific growth and motion, enabling maternity patterns that evolve with individual pregnancy trajectories.
Pediatric & infant clothing
4D scans of crawling and walking reveal where garments restrict or assist movement, guiding development-safe baby and kids clothing.