Motion Tracking — WAVGEN AV Universe

WAVGEN · Motion Tracking

MoCap
IK
Velocity
Gesture
BVH
Artists
Glossary

FRAME 0000

JOINTS 23

LATENCY 8ms

MODE OPTICAL

AV Universe · Realtime Movement Intelligence

Motion
Tracking

Where the body becomes data — skeletal structures flowing through space, every joint a coordinate, every gesture a signal.

Section 02 · Capture Technology

Motion Capture Systems

Motion capture spans optical marker-based systems, inertial measurement units, markerless video-based capture, and electromagnetic field tracking. Each trades accuracy, occlusion handling, cost, and portability differently.

Comparative System Visualization — Click System to Inspect

Optical Passive

Retroreflective markers tracked by infrared cameras. Sub-millimeter accuracy in controlled studio environments.

Accuracy: 0.1mm · Markers: 50+ · Cameras: 8–24

Inertial (IMU)

Accelerometers and gyroscopes per body segment. Portable, no camera rig — but accumulates drift over time.

Accuracy: 1–3° · Sensors: 17 · Drift: ~1°/min

Markerless Video

ML-based pose estimation from RGB cameras. No setup, natural performance — constrained by occlusion and depth ambiguity.

Accuracy: 5–20mm · Cameras: 1–6 · RT: yes

RGB-D / Depth

Structured light or time-of-flight depth sensors combined with RGB provide 3D joint estimation without markers.

Accuracy: 3–10mm · Range: 0.5–8m · FPS: 30–90

Section 03 · Joint Solving

Inverse Kinematics

IK solvers compute joint angles required to place an end-effector (hand, foot) at a target position. FABRIK (Forward And Backward Reaching IK) iteratively adjusts joint chains toward targets while maintaining bone lengths. Move cursor to drag the end-effector.

FABRIK IK Solver — Move Cursor to Control End-Effector

Chain Parameters

Algorithm: FABRIK

Iterations: 10
Convergence: 0.001
Joint Count: 7

Bone 1: —°
Bone 2: —°
Bone 3: —°
Bone 4: —°
Bone 5: —°
Bone 6: —°

Reach: —px
Error: —px

Iterations 10

Stiffness 0.0

Section 04 · Biomechanics

Velocity & Acceleration Analysis

Biomechanical analysis of joint trajectories reveals velocity profiles, acceleration peaks, and energy transfer patterns. Real-time differentiation of position data yields per-joint velocity; a second derivative yields acceleration, highlighting impact events and movement initiation.

Playback Speed 1.0x

Smoothing Window 8

Velocity Profile — Wrist Joint

Acceleration — Impact Events

Section 05 · Recognition

Gesture Recognition

Gesture classifiers map sequences of joint positions to discrete gesture labels using DTW (Dynamic Time Warping), HMMs, or sequence-to-label neural networks. Real-time systems maintain a sliding window of recent poses and emit class probabilities continuously.

Window Length 40

Real-Time Classification — Move Cursor over Canvas

Section 06 · Data Interchange

BVH & Mocap Formats

BVH (Biovision Hierarchy) encodes skeleton hierarchy and keyframe channel data in a plain-text format readable by most DCC tools. C3D stores marker cloud and analog data in binary. FBX wraps skeleton, animation, and mesh in a proprietary Autodesk container widely supported across pipelines.

FRAME: 000 / 240 FORMAT: BVH

BVH Skeleton Playback — 23 Joints · 240 Frames

Section 07 · Feature Tracking

Sparse Feature Tracking

Lucas-Kanade sparse optical flow tracks a set of detected interest points (Harris corners, Shi-Tomasi) frame-to-frame using local gradient descent. The resulting per-point velocity vectors reveal rigid body motion, camera shake, and scene dynamics.

Point Count 40

Trail Length 30

Lucas-Kanade Tracker — Move Cursor to Drive Motion

Section 08 · Practitioners

Artists & Researchers

Motion tracking intersects dance, sports science, film VFX, and interactive art — enabling practitioners to make movement computational, expressive, and analytical.

Section 09 · Reference

Glossary

Core terminology spanning motion capture systems, skeletal animation, biomechanical analysis, and real-time tracking pipelines.

WAVGEN AV Universe · Motion Tracking · Realtime Movement Intelligence · 2024