A: Swing Catalyst Motion Capture (MoCap) is a markerless 2D motion capture technology that analyzes golf movements using cameras and AI algorithms. It requires no physical markers or sensors.

A: Swing Catalyst Motion Capture offers the following capabilities:

• Intelligent person detection
• Precise 2D pose estimation with up to 26 key points
• Near real-time pose visualization with minimal latency
• Skeleton overlay with markers for joints
• Visualization of COM (Center of Mass)
• Joint angle measurement
• Custom joint-to-joint line visualization
• Movement path tracing for specific joints

• Active Swing Catalyst Pro or Pro+ subscription (version 25.1 and newer)
• Compatible computer meeting recommended specifications
• One or more supported cameras

A: The Swing Catalyst Motion Capture feature is available in version 25.1.

• Up to 4 simultaneous cameras (excluding top-down view)
• Compatible angles:
  • Face-on right
  • Face-on left
  • Down the line
• Note: Top-down and freehand angles are not supported

A: For multi-camera or high FPS setups:

• Newer generation NVIDIA cards with more VRAM (8-12GB or more) generally provides better performance, especially with TensorRT or CUDA at higher batch sizes. The GPU models listed below also include the Ti, Ti Super and Super variants from NVIDIA.
• NVIDIA RTX 3070, 3080, 3090
• NVIDIA RTX 4060, 4070, 4080, 4090
• NVIDIA RTX 5060, 5070, 5080, 5090

• Execution Provider Priority (NVIDIA): TensorRT → CUDA → DirectML
• For AMD or Intel GPU users please use DirectML
• Batch size: Depends on how many cameras and dedicated GPU memory your computer has, see our support article
• Optional optimizations:
  • Enable Nano detection model (if not default)
  • FP16 for an additional performance boost (this will reduce the accuracy of the model, only applicable to TensorRT at this time)

• Complete the camera calibration process
• Ensure proper camera alignment and positioning (e.g ensuring the subject is centered and the cameras are level)
• Maintain consistent, well-distributed lighting

• Verify execution provider selection (TensorRT recommended for NVIDIA cards)
• Initial TensorRT optimization may take several minutes; subsequent recordings will be faster
• Consider enabling the Nano detection model for better performance

A: Yes. In general rapid movements and complex poses may occasionally result in tracking errors. Functionality is currently still being developed and is subject to change.

1. Ensure your software is up to date (version 25.1 or newer)
2. Locate the MoCap button in the top menu bar
3. Click to toggle the feature on/off
4. Note: The MoCap feature can be turned on/off independently for capture or playback.

Please see our user guide for more information.

A: Any live streaming camera supported by Swing Catalyst works. See our supported cameras article.

A: Processing times are highly hardware dependent.

Certain things such as the type of graphics card, batch size, execution provider can make a tremendous difference in processing speeds. With MoCap enabled we estimate it will take anywhere from 30-50% longer for a recording to process (the time it takes from a recording is triggered until it's ready to be opened) with MoCap enabled than with it disabled.

If you need to make a lot of recordings sequentially you can turn off MoCap, make your recordings, then when you open the recordings later turn on MoCap then run the re-processing step.

A: It's likely person detection is failing in this scenario.

1. Try to toggle capture mode on/off
2. Try to toggle MoCap off/on
3. Try to change the detection model from Nano to Medium, see our settings for more details
4. Make sure your camera(s) are setup correctly according to our recommendations