System Overview
YoomTrack Performance Capture is designed to track human movement in 3D space without the use of physical markers attached to the subject. It utilizes advanced computer vision algorithms, multiple cameras, and machine learning techniques to accurately capture and reconstruct full-body, hands, and face movements. It is suitable for applications such as film production, animation, virtual performances, virtual reality, gaming, biomechanics, and sports performance analysis.
KEY FEATURES
No Physical Markers Required: Utilizes cameras and machine learning algorithms to capture movement without requiring markers or suits.
Multi-Camera Support: Supports multiple camera setup for capturing full-body, hands and face motion from different angles.
Real-Time Tracking: Capable of providing real-time motion data stream for quick review and selection processes.
High Accuracy and Precision: Achieves high levels of accuracy in motion tracking for both basic and complex movements using advanced automated cleanup algorithms
Natural Performance Capture: Ideal for capturing unencumbered, natural movement in environments without marker setups.
Environment Flexibility: Can operate in a variety of environments, from controlled studios to more complex, dynamic locations.
Multiangle reference cameras: Utilizes all setup cameras as potential reference cameras to streamline selection
Data Output Formats: Common output formats include FBX and other industry standard formats, allowing easy integration with 3D animation, game engines, and other software.
SYSTEM COMPONENTS
CAMERAS
- Type: Vision cameras, Optical Motion Capture cameras.
- Resolution: Typically 5MPX or higher.
- Exposure Time: 1-3 ms
- Number of Cameras: Typically requires 14 to 40 cameras placed around the capture space for optimal coverage (depending on number of performers and stage size)
SOFTWARE SUITE
- Stage Manager: capture stage and production day management tool for efficient execution of production days.
- Calibration Tools: Auto-calibration of the system to ensure the accurate placement of captured data within the virtual environment.
- Real-Time Data Processing and Review: Enables live output of character animation or performance capture directly into digital assets.
- Production Manager: data management and process management tool for performance capture processing, automated cleaning and team collaboration.
- Grid Manager: Manages the allocated processing resources.
- Performance Capture Solve and Automated Cleanup: Advanced algorithms that track body joints, post-processing of raw camera data into skeletons and motion curves.
- Integration Support: Compatible with 3D software like Maya, Blender, Unity, Unreal Engine, etc.
- Accuracy Level: Typically within 1-5 cm for joint position, depending on the quality of the camera system and the number of cameras used.
- Size of Capture Area: From small (3x3m) for basic applications to large (15x15m or more) for full-body or ensemble captures.
- Space Requirements: Works best in a studio environment with ample lighting and a clear, unobstructed view of the subject. Systems can also work in natural environments but require adequate space for camera placement.
CAMERA SPECS
- Resolution: 5MPX cameras or higher
- Exposure Time: 1-3 m
- Lens Type: Wide-angle lenses (12mm Computar lenses)
- Synchronization: Time-code or genlock for camera sync.
SOFTWARE AND PROCESSING
- Data Output: FBX and other industry standard animation file formats.
- Processing Speed: Real-time tracking with minimal latency. Automated cleaning speed at 1 minute per second of cleaning at standard configuration (4 processing agents, scalable with amount of agents).
- Platform Compatibility: Windows and Linux (for processing agents).
- Integration: Compatible with 3D animation tools (e.g., Maya, 3ds Max), game engines (Unreal Engine, Unity), and VFX software (Houdini, Nuke).
Internet Connection Wired Support: Support for Ethernet and COEX for connecting cameras and servers
Real-Time Feedback: Immediate data visualization for fine-tuning and performance adjustments.
Multi-Actor Tracking: Capability to track multiple actors simultaneously (limited only by space and camera configuration).
Occlusion Handling: Predictive Intelligent tracking for partial body occlusions (e.g., behind objects or other people).
Calibration System: Automatic system calibration to maintain accuracy over time.
High-Speed Capture Mode: For high-speed movements (e.g., sports performance, martial arts).
Hardware Requirements (for processing agents):
- Processing Power: Intel® Core™ i9-13900K Processor, 64 GB (4 x 16 GB) DDR5
- 4800 UDIMM Memory
- Graphics Card: GFX NVIDIA RTX 4090 24GB GDDR6X
- Storage: 512 GB HP Z Turbo Drive M.2 SSD
Software Requirements:
- OS: Windows 10
- Optional Additional Software: Compatible 3D modeling, animation, or game engine software (e.g., Unreal Engine, Unity, Autodesk Maya, Motion Builder).
