ARKit TrueDepth Mesh

1220 3D points + 52 blendshapes, 60 FPS

FACS Action Units

Expression coding standard with deep analysis

Deception Markers

6 key deception markers: asymmetry, micro-tension, fear of exposure, concealed satisfaction, reaction delay (>300ms), emotion/word mismatch

Temporal Analysis

Trends, anomalies, cumulative stress

Cross-Signal Fusion

Multimodal consensus of up to 13 channels

Voice DSP

Pitch (F0), jitter, tremor, pauses, response latency — from the live audio stream

rPPG (Remote PPG)

Heart rate and SpO2 from the facial skin video signal — fully contactless

Eye Tracking + Pupillometry

Gaze direction, blink patterns, pupil dynamics — frame by frame

Edge Processing

In-browser processing — data never leaves the device

WebRTC + WebAudio

Native camera and microphone access

Real-time Pipeline

Capture → Landmarks → AU → Emotion in <100ms

Individual Baseline

Each person’s norm comes from their own calibration, not someone else’s statistics

AES-256 Encryption

Encrypted reports, operator-only access

Apple Watch Direct

Direct stream from the wrist: HR, HRV, breathing, tremor — no cloud

Red Rabbit Ring

Our own sensor: GSR, heart rate, SpO2, tremor — channels no consumer gadget provides

Ephemeral by Design

On-the-fly analysis: video is never stored, only the assessment remains

Patent Pending

98 claims across two USPTO filings: multimodal scoring, wearables, ephemeral architecture