Skip to main content

Multicam fundamentals

Multi-camera motion capture is the foundation of professional-grade motion capture systems. This guide explains the key concepts and setup requirements for multi-camera workflows.

What is multi-camera motion capture?​

Multi-camera motion capture uses multiple synchronized cameras to capture motion from different angles simultaneously. This approach provides:

  • Better Accuracy: Multiple viewpoints reduce occlusion and improve precision
  • 3D Reconstruction: True 3D motion capture data instead of 2D projections
  • Professional Quality: Studio-grade results suitable for production
  • Complex Scenes: Ability to handle multiple subjects and complex movements

Key components​

Cameras​

Multiple cameras positioned around the capture volume to provide overlapping viewpoints.

Calibration​

The process of determining the spatial relationship between cameras in 3D space.

Rigs​

3D skeleton templates that define the skeletal structure for motion capture data mapping.

Lenses​

Camera lens specifications that affect how the 3D world is projected onto 2D images.

Camera setup​

Camera positioning​

For optimal multi-camera capture:

  1. Capture Volume: Define the 3D space where motion will occur
  2. Camera Placement: Position cameras around the volume with overlapping views
  3. Angle Coverage: Ensure all parts of the subject are visible from multiple angles
  4. Distance: Balance between coverage and resolution

4-Camera Setup (Minimum)​

    Camera 1
        |
        |
Camera 2 -+- Camera 3
        |
        |
    Camera 4

8-Camera Setup (Professional)​

Camera 1    Camera 2
    |         |
    |         |
Camera 3 -+- Camera 4
    |         |
    |         |
Camera 5    Camera 6
    |         |
    |         |
Camera 7    Camera 8

Calibration​

What is calibration?​

Calibration determines the position, orientation, and lens parameters of each camera relative to a common 3D coordinate system.

Calibration process​

  1. Calibration Object: Use a known-size calibration object (checkerboard, wand, etc.)
  2. Multiple Views: Capture the calibration object from all camera angles
  3. Algorithm Processing: Software calculates camera positions and parameters
  4. Validation: Verify calibration accuracy with test captures

Calibration data​

Calibration includes:

{
  "cameras": [
    {
      "id": "cam_1",
      "position": [x, y, z],
      "rotation": [rx, ry, rz],
      "intrinsics": {
        "focal_length": [fx, fy],
        "principal_point": [cx, cy],
        "distortion": [k1, k2, p1, p2]
      }
    }
  ],
  "coordinate_system": {
    "origin": [0, 0, 0],
    "units": "meters"
  }
}

Calibration best practices​

  • Stable Setup: Ensure cameras don't move after calibration
  • Good Coverage: Calibration object should be visible from all cameras
  • Multiple Positions: Capture calibration object in different positions
  • Regular Recalibration: Recalibrate if cameras are moved or bumped

Rigs​

What is a rig?​

A rig is a 3D skeleton template that defines the skeletal structure for motion capture data. It specifies:

  • Joint Hierarchy: Parent-child relationships between bones
  • Joint Names: Standardized naming for skeletal joints
  • Default Pose: Rest pose or T-pose for the skeleton
  • Scale: Physical dimensions of the skeleton

Rig types​

Humanoid Rigs​

Standard human skeleton with common joint names:

  • Head, Neck, Spine
  • Shoulders, Elbows, Wrists
  • Hips, Knees, Ankles

Custom Rigs​

Specialized skeletons for specific use cases:

  • Animals (horses, dogs, etc.)
  • Characters with non-standard proportions
  • Mechanical or robotic structures

Rig Definition​

{
  "name": "humanoid_standard",
  "joints": [
    {
      "name": "Hips",
      "parent": null,
      "position": [0, 0, 0]
    },
    {
      "name": "Spine",
      "parent": "Hips",
      "position": [0, 1, 0]
    },
    {
      "name": "LeftHip",
      "parent": "Hips",
      "position": [-0.1, 0, 0]
    }
  ],
  "default_pose": "T-pose",
  "units": "meters"
}

Lenses​

Lens parameters​

Camera lenses affect how 3D space is projected onto 2D images:

  • Focal Length: Distance from lens to image sensor
  • Field of View: Angular extent of the scene captured
  • Distortion: Lens distortion that needs correction
  • Aperture: Controls depth of field and light intake

Lens calibration​

Lens calibration determines:

{
  "focal_length": [fx, fy],
  "principal_point": [cx, cy],
  "distortion_coefficients": [k1, k2, p1, p2],
  "image_size": [width, height]
}

Multi-camera workflow​

1. Setup Phase​

  • Position cameras around capture volume
  • Ensure stable mounting and power
  • Verify camera synchronization

2. Calibration Phase​

  • Perform camera calibration
  • Validate calibration accuracy
  • Save calibration data

3. Capture Phase​

  • Record synchronized video from all cameras
  • Ensure subject stays within capture volume
  • Maintain good lighting conditions

4. Processing Phase​

  • Upload all video files
  • Provide calibration data
  • Select multi-camera model (m1 or m2)
  • Process and generate motion capture data

Best practices​

Camera setup​

  • Stable Mounting: Use tripods or wall mounts
  • Power Management: Ensure reliable power supply
  • Network Sync: Use network synchronization for precise timing
  • Lighting: Provide consistent, even lighting

Calibration​

  • High-Quality Object: Use precise calibration objects
  • Multiple Views: Capture from various angles
  • Validation: Test calibration with known movements
  • Documentation: Keep calibration data organized

Capture​

  • Volume Awareness: Keep subjects within calibrated volume
  • Lighting Consistency: Maintain stable lighting throughout
  • Camera Monitoring: Monitor all cameras during capture
  • Backup Recording: Always have backup recordings

Model selection​

M1 Model​

  • Use Case: Standard multi-camera setups
  • Accuracy: Good accuracy for most applications
  • Speed: Faster processing than m2
  • Requirements: 2-8 cameras, basic calibration

M2 Model​

  • Use Case: Complex scenes, high accuracy requirements
  • Accuracy: Highest accuracy available
  • Speed: Slower processing than m1
  • Requirements: 4+ cameras, precise calibration

Next steps​