Point Cloud Soup Recorder

Reconstruct high-resolution point clouds from Hammerhead's PointCloudSoup messages, and record as PLY files.

Build

mkdir build
cd build
cmake ..
cmake --build . --config Release

Usage

# Linux
./point_cloud_soup_recorder [OPTIONS] [hammerhead_ip] [output_directory]

# Windows
./Release/point_cloud_soup_recorder.exe [OPTIONS] [hammerhead_ip] [output_directory]

Options

  • -w, --wait-for-scheduler: Enable scheduler synchronization with Hammerhead. When enabled, the recorder will wait for Hammerhead to request the next frame before continuing, ensuring frame-by-frame synchronization.
  • -h, --help: Display usage information

Parameters

  • hammerhead_ip: IP address of the device running Hammerhead (default: 127.0.0.1)
  • output_directory: Directory to save PLY files (default: point_clouds folder)

Examples

# Record point clouds from local device (default)
./point_cloud_soup_recorder

# Record point clouds from local device with custom output directory
./point_cloud_soup_recorder 127.0.0.1 /tmp/ply_output

# Record point clouds from remote device with custom output directory
./point_cloud_soup_recorder 10.10.1.10 /tmp/ply_output

# Record with scheduler synchronization (frame-by-frame)
./point_cloud_soup_recorder -w 10.10.1.10 /tmp/ply_output
./point_cloud_soup_recorder --wait-for-scheduler 10.10.1.10 /tmp/ply_output

Scheduler Synchronization Workflow

When using the -w flag for frame-by-frame synchronization:

  1. Configure Hammerhead - Edit master_config.ini: 

    wait_for_scheduler = 1

  2. Start the recorder first: 

    ./point_cloud_soup_recorder -w

  3. Start Hammerhead - The recorder will now control frame processing timing

This ensures proper ZMQ connection establishment.

Output

  • Format: PLY files (.ply)
  • Location: point_clouds folder
  • Naming: Sequential numbering based on received messages

Features

  • High-performance C++ implementation for optimal processing speed
  • Subscribe to PointCloudSoup messages from Hammerhead
  • Reconstruct full point clouds from compact soup representation
  • Generate PLY files compatible with CloudCompare and other tools
  • Handle high-resolution point clouds efficiently with minimal memory usage
  • Optional scheduler synchronization for frame-by-frame processing with Hammerhead

PointCloudSoup Format

Nodar generates extremely high-resolution point clouds that require efficient network transmission. The PointCloudSoup format:

  • Provides a compact representation of point cloud data
  • Allows reconstruction of full point clouds on client machines
  • Reduces network bandwidth requirements significantly
  • Maintains high fidelity of 3D spatial information

The point_cloud_soup.hpp header provides the class and functions for reading and writing PointCloudSoup data.

Troubleshooting

  • No point clouds generated: Check IP address and ensure Hammerhead is running
  • Connection hanging: ZMQ will wait indefinitely for connection - verify network connectivity
  • Large file sizes: Point clouds are high resolution - ensure adequate storage space

Press Ctrl+C to stop recording.