Motion Capture / MoCap

Definition

Motion Capture (or MoCap) is the technology for digitally capturing and recording body movements in real-time. An actor or an animated character wears a special suit with reflective markers that are tracked by infrared-sensitive cameras. The captured positions are converted into digital skeletal data, which is then applied to CGI characters.

Motion capture is essential today for Hollywood blockbusters featuring digital characters like "Avatar," "The Jungle Book," or the Marvel films. The process enables photorealistic motion animation that would be impossible to capture by hand.

Types of Motion Capture

1. Optical Marker-Based MoCap (Standard)

How it Works:

• Reflective markers (12-16mm diameter) are attached to the body
• Infrared cameras capture the 3D position of each marker
• Triangulation calculates precise skeletal positions
• Real-time calculation allows for live preview

Equipment:

• 12-32 specialized infrared cameras (OptiTrack, Vicon, Xsens)
• Reflective marker sets
• Special suits with marker pockets
• Real-time tracking software

Advantages:

• Highest accuracy (sub-millimeter)
• Multiple actors possible simultaneously
• Very fast processing
• Unlimited movement area possible

Disadvantages:

• Expensive (Stage rental: €8-15K/day)
• Marker occlusion (obscuration) is problematic
• Special suit is uncomfortable
• Calibration is complex

2. Inertial MoCap (IMU-Based)

How it Works:

• Accelerometers on each joint
• No external camera needed
• Wireless data transmission
• Lower latency possible

Examples: Xsens MVN, OptiTrack Geno

Advantages:

• Outdoor captures possible
• No camera setup required
• Faster to deploy
• Cheaper than optical systems

Disadvantages:

• Drift and noise over time
• Less precise than optical
• Calibration before each session
• Expensive per suit (€50-70K)

3. Markerless / AI-Based MoCap

How it Works:

• Deep learning algorithms recognize body joints from video
• No markers or special hardware needed
• Real-time processing on standard GPUs
• Increasingly available (OpenPose, MediaPipe, RunwayML)

Advantages:

• Inexpensive (Software costs €100-500/month)
• Quick to implement
• No special equipment
• Indoor & Outdoor

Disadvantages:

• Less precision (±5-10cm error)
• Only one person per take
• Weak with fast movements
• Data post-processing necessary

4. Real-Time / Live MoCap (Streaming)

How it Works:

• Real-time tracking is fed directly into a 3D engine
• Actor sees their digital alter-ego on a live monitor
• Interactive performance possible
• Used in Virtual Production (LED stages)

Examples: "The Mandalorian," "Fortnite Performance Capture"

Advantages:

• Live feedback for actors
• Director can make adjustments in real-time
• Reduces rework
• Real-time previsualization

Disadvantages:

• Extremely expensive (€100K-200K/day)
• Technically complex
• Specialized talent required
• Limited technical error tolerance

Marker Placement: Standard Skeleton

A standard MoCap skeleton typically has 40-70 markers:

Head:
├── Crown
├── Forehead
├── Back_Head
└── Neck

Spine:
├── Spine_1 (lower)
├── Spine_2 (middle)
├── Spine_3 (upper)
└── Clavicle_L/R (collarbones)

Left Arm:
├── Shoulder_L
├── Elbow_L
├── Wrist_L
├── Hand_L
└── Finger_L [1-5]

Right Arm:
└── (identical)

Pelvis:
├── LHIP (left hip)
├── RHIP (right hip)
└── Pelvis_Back (rear)

Left Leg:
├── Knee_L
├── Ankle_L
├── Toe_L
└── Heel_L

Right Leg:
└── (identical)

MoCap Workflow

Phase 1: Pre-Production

Before the Capture Session:

• Scene planning and blocking
• Marker placement definition
• Camera setup and calibration
• Suit fitting & size selection
• Actor briefing

Phase 2: Capture (in Studio)

Preparation (30 min):
├── Actor puts on MoCap suit (30kg with equipment)
├── Marker placement and verification
└── Camera calibration run (T-Pose & A-Pose)

Recording (4-6 hours):
├── Record takes
├── Real-time QC check
├── Retake if marker occlusion occurs
└── T-Poses between takes for reference

Post-Capture (30 min):
├── Data validation
├── File transfer and backup
└── Equipment cleaning

Phase 3: Post-Processing (2-4 weeks)

Raw Capture Data
├── Marker gap-filling (interpolation for missing frames)
├── Jitter reduction & smoothing
├── Skeleton fitting (marker → skeleton conversion)
├── Scale & T-Pose normalization
├── Motion graph creation
└── FBX/EXR export for animation

Technical Specifications

Optical Tracking System (Industry Standard)

Accuracy: ±2-5mm RMS Error
Latency: 2-4 frames (at 24fps = 83-166ms)
Capture Framerate: 120-240fps (for downsampling to 24fps)
Workspace: 4m x 4m to 20m x 20m (any size with array)
Number of Cameras: 12-32 cameras typical
Refresh Rate: 120Hz or 240Hz

Data Format & Size

An 8-hour session with 50 markers at 120fps:
├── Raw Data: ~50-80 GB (proprietary format)
├── Skeleton Data: ~2-5 GB (FBX/BVH)
├── Motion Graph: ~500MB-1GB
└── Archive Backup: 150-200 GB (redundant)

Problems & Solutions in MoCap

Problem 1: Marker Occlusion

What: Markers are obscured by body parts, tracking system loses position

Solutions:

• Marker gap-filling by software (interpolation)
• Increase physical distance between markers
• Higher camera count (redundant sightlines)
• Manually clean up problem areas

Cost for Cleanup: +30-50% of post-production time

Problem 2: Jitter & Noise

What: Markers "jitter" due to camera noise or reflections

Solutions:

• Software-based jitter reduction (Butterworth filter)
• Manual keyframe correction
• Higher capture framerate for downsampling
• Better marker quality (reflective properties)

Problem 3: Shoulder Pop / Gimbal Lock

What: Unnatural shoulder rotations due to mathematical singularities

Solutions:

• Quaternion-based rotation (instead of Euler angles)
• Solver constraints in the skeleton system
• Manual hand animation for critical frames
• Higher-order interpolation

Problem 4: Finger Movements

What: 5 fingers per hand are difficult to track (high marker density)

Solutions:

• Specialized hand-tracking cameras (separate)
• Gloves with finger markers
• Semi-automatic hand animation
• Often manually post-processed (80% of shots)

MoCap vs. Hand Animation

Actor Performance in MoCap

What Works:

• Large-scale, clear movements
• Body language & posture
• Emotional expression through movement
• Interaction with other MoCap actors
• Dynamic action sequences

What is Difficult:

• Subtle micro-movements
• Finger gestures
• Eye contact (filmed separately)
• Clothing interaction
• Realistic object grabbing

Future of Motion Capture

Current Trends:

• Real-time AI-assisted marker occlusion handling
• Markerless systems are continuously improving (RunwayML, OpenPose 2.0)
• Live MoCap in streaming productions
• Hybrid approaches (optical + IMU combined)
• Cloud-based post-processing

See Also

• CGI – Digital Characters & Environments
• Animation – Creating Movement Digitally
• VFX Supervisor – Quality Control
• Virtual Production – Real-Time MoCap on Set