Open Video Tracker User Guide

Introduction

Open Video Tracker is a Blender addon that automates the process of converting video footage into 3D point clouds and camera tracks using photogrammetry techniques. The addon integrates FFmpeg, COLMAP, and GLOMAP to provide a streamlined workflow for video tracking and 3D reconstruction.

Installation

Step 1: Download the Addon

1. Download Open Video Tracker addon file (only avalible on windows for now!)
   • Download cuda version if you have a Rtx Nvidia GPU
   • Else download normal version
2. Locate the open_video_tracker folder containing the addon files

Step 2: Install in Blender

1. Open Blender
2. Go to Edit → Preferences
3. Select the Add-ons tab
4. Click Addon Settings [^]
5. Install from disk
6. Navigate to and select the open_video_tracker folder
7. Click Install From Disk

Step 4: Verify Installation

1. Switch to the 3D Viewport
2. Look for the "Open Video Tracker" tab in the right sidebar
3. If visible, the installation was successful

User Interface Pverview

Imported Data Preview

Main Panel

The main interface is located in the 3D Viewport sidebar under the "Open Video Tracker" tab.

Video Selection Section

• Video Path: File browser to select input video file
• Video Information: Displays frame rate, resolution, and bitrate (populated automatically)

Frame Extraction Settings

• Quality: Preset quality levels for frame extraction (Native, High, Balanced, Low, Lowest)

Feature Extraction Settings

• Max Image Size: Maximum dimension for feature extraction (default: 2000px)
• Use GPU: Enable GPU acceleration for feature extraction (if you have cuda version)
• Camera Model: Camera distortion model (Simple Radial recommended for most cases)
• Max Num Features: Maximum features to extract per image (default: 8192)

Sequential Matching Settings

• Overlap: Number of overlapping frames for sequential matching (default: 10)

Reconstruction Settings

• Max Tracks: Maximum tracks per image (default: 1000)
• Constraint Type: Balance between point and camera constraints
• Advanced Options: Additional GLOMAP parameters (epipolar error, iterations)

Execution Controls

• Track Video: Button to start the processing pipeline
• Progress: Real-time progress indicator during processing

Import Options Panel

Located in the collapsible "Import Options" panel:

Camera Import Settings

• Import Cameras: Enable camera track import
• Camera Extent: Size of camera visualization
• Add Background Images: Include original frames as background
• Add Image Planes: Create image plane objects
• Animation Options: Camera motion animation settings

Point Import Settings

• Import Points: Enable point cloud import
• Point Cloud Display: Sparsity and rendering options
• GPU Rendering: Hardware-accelerated point display
• Mesh Options: Convert points to mesh objects

Workflow

Step 1: Prepare Your Video

1. Ensure your video file is in a supported format (MP4, AVI, MOV, MKV, WMV, FLV, WebM)
2. For best results:
   • Use high-quality, stable footage
   • Avoid excessive camera movement or shaky footage
   • Ensure good lighting and contrast
   • Keep video length reasonable (start with short clips for testing)

Step 2: Configure Processing Settings

1. Video Quality: Choose based on your needs and hardware

   • Native/High: Best quality, largest files, slowest processing
   • Balanced: Good quality/performance balance
   • Low/Lowest: Fast processing, smaller files

2. Feature Extraction:

   • Increase Max Image Size for high-resolution videos
   • Enable GPU if available for faster processing
   • Choose appropriate camera model based on your camera type

3. Reconstruction:

   • Adjust Max Tracks based on scene complexity
   • Use Balanced constraint type for most scenarios

Step 3: Run the Pipeline

1. Click "Track Video" to start processing
2. Monitor progress in the progress bar
3. Processing time varies based on video length and settings
4. Do not close Blender during processing

Step 4: Import Results

After processing completes, the results are automatically imported. The addon creates:

• Camera objects with animation data
• Point cloud data
• Background images (if enabled)

Configuration Options

Camera Models

Choose the appropriate camera model based on your camera type:

• Simple Radial: Most consumer cameras (smartphones, DSLRs)
• Simple Radial Fisheye: Action cameras (GoPro)
• OpenCV: Computer vision cameras
• Full OpenCV: Complex distortion scenarios

Constraint Types

• Points Only: Use when camera positions are unreliable
• Cameras Only: Use when point features are sparse
• Balanced: Recommended for most scenarios
• Points and Cameras: Maximum constraints, slower processing

Running the Pipeline

Progress

• Each step is logged to the Blender console
• Processing can be monitored via Window → Toggle System Console

Expected Output

The addon uses same dir as th blend file and creates following directory structure:

{your_blend_file_directory}/
└── video_tracking/
    └── {video_name}/
        ├── images/          # Extracted frames
        ├── sparse/          # Reconstruction data
        │   ├── 0/          # Model files
        │   └── cameras.txt # Camera data
        │   └── images.txt  # Image data
        │   └── points3D.txt # Point cloud data
        └── database.db     # COLMAP database

Importing Results

Import Options

Camera Import

• Camera Extent: Visual size of camera objects
• Background Images: Show original frames
• Image Planes: Create textured planes
• Animation: Convert camera motion to keyframes

Point Cloud Import

• Display Sparsity: Reduce point density for performance
• GPU Rendering: Hardware-accelerated display
• Mesh Conversion: Convert points to mesh objects

Post-Processing

Point Cloud Processing

• Comming soon

Best Practices

Video Preparation

• Resolution: Higher resolution provides better tracking but increases processing time
• Frame Rate: 24-30 fps is optimal; higher rates may not improve tracking significantly
• Stabilization: Use stabilized footage when possible
• Lighting: Ensure consistent, even lighting
• Motion: Smooth camera movement works better than jerky motion

Processing Optimization

• Start Small: Test with short video clips first
• GPU Usage: Enable GPU acceleration when available
• Quality Settings: Use Balanced quality for initial tests
• Memory Management: Monitor RAM usage with large videos

Scene Considerations

• Feature-Rich: Scenes with lots of texture and detail track better
• Motion Blur: Minimize motion blur for better feature detection
• Scale: Include objects of known size for scale reference
• Overlap: Ensure sufficient scene overlap between frames

Quick Fixes

• Processing Fails: Check executable paths in preferences
• Out of Memory: Reduce Max Image Size or use lower quality
• Poor Tracking: Adjust camera model or increase Max Features
• Slow Processing: Enable GPU or reduce video resolution

Credits

Blender — 3D creation suite (blender.org)
FFmpeg — Video processing toolkit (ffmpeg.org)
COLMAP — Structure-from-Motion and MVS (colmap.github.io)
GLOMAP — Global incremental SfM (github.com/colmap/glomap)
Blender Photogrammetry Importer (github.com/SBCV/Blender-Addon-Photogrammetry-Importer)
Thanks to the open‑source community and contributors

Logos and trademarks are the property of their respective owners.