Camera Interface Implementation

Guides the implementation of camera hardware interfaces using the ataraxis-video-system library. This skill focuses on the low-level hardware integration patterns applicable to any acquisition system.

When to Use This Skill

Use this skill when:

• Adding camera support to an acquisition system
• Troubleshooting camera connectivity issues
• Verifying camera configuration before runtime
• Testing cameras with interactive acquisition
• Understanding the VideoSystem API

For system-specific integration (modifying sl-shared-assets configuration, integrating into mesoscope-vr), use the

/modifying-mesoscope-vr-system

Verification Requirements

Before writing any camera code, verify the current state of dependent libraries.

Step 0: Version Verification

Follow the Cross-Referenced Library Verification procedure in

CLAUDE.md

1. Check local ataraxis-video-system version against GitHub
2. If version mismatch exists, ask the user how to proceed
3. Use the verified source for API reference

Step 1: Content Verification

../ataraxis-video-system/README.md

../ataraxis-video-system/src/ataraxis_video_system/__init__.py

../ataraxis-video-system/src/ataraxis_video_system/video_system.py

pyproject.toml

Step 2: Hardware Verification

Before implementing camera code, verify cameras are connected and accessible using MCP tools.

The ataraxis-video-system library provides an MCP server for camera discovery. Start the server with:

axvs mcp

Verification workflow:

1. Check runtime requirements:

   check_runtime_requirements()

   - Verify FFMPEG and GPU availability
2. Check CTI status (if using Harvesters):

   get_cti_status()

   - Verify GenTL Producer is configured
3. Discover cameras:

   list_cameras()

   - Verify expected cameras are detected with correct indices

Expected output from

list_cameras()

OpenCV Cameras:
  Index 0: 1920x1080 @ 30fps
  Index 1: 640x480 @ 30fps

Harvesters Cameras:
  Index 0: Allied Vision Mako G-040B (1936x1216)

If cameras are not detected:

• Check physical USB/GigE connections
• Verify camera drivers are installed
• For Harvesters: ensure CTI file is configured (

  get_cti_status()

  )
• Check for port conflicts with other applications

Do not proceed with implementation until expected cameras are verified.

Camera Discovery

Use the ataraxis-video-system MCP tools for camera discovery. These tools provide programmatic access to connected cameras.

MCP Tools Available

list_cameras

get_cti_status

set_cti_file

check_runtime_requirements

Discovery Workflow

1. Check runtime requirements: Verify FFMPEG and GPU availability
2. Check CTI status: Ensure GenTL Producer is configured for Harvesters cameras
3. List cameras: Discover all connected cameras with their indices and properties
4. Record camera indices: Note the indices for configuration files

Interactive Camera Testing

Test cameras using the MCP video session tools before integrating into the acquisition system.

Test Session Workflow

1. Start video session: Initialize camera with test parameters
2. Verify preview: Check that frames are being acquired and displayed
3. Start frame saving: Test recording to temporary directory
4. Stop session: Clean up resources

MCP Session Tools

start_video_session

stop_video_session

start_frame_saving

stop_frame_saving

get_session_status

Test Parameters

For testing, use conservative parameters with a temporary directory:

# Test configuration
output_directory = None          # Use temp directory (or specify path for manual verification)
camera_interface = "harvesters"  # or "opencv"
camera_index = 0                 # From discovery
display_frame_rate = 15          # Low for testing
video_encoder = "h264"           # CPU encoding for compatibility
quantization_parameter = 25      # Moderate quality

The MCP tools use a temporary directory by default for test recordings. Specify an output directory only if the user wants to manually verify the recorded video file.

VideoSystem API Reference

See CAMERA_INTERFACE_GUIDE.md for the complete API reference including:

• VideoSystem constructor parameters
• Enumeration values (CameraInterfaces, VideoEncoders, EncoderSpeedPresets)
• Lifecycle methods (start, stop, start_frame_saving, stop_frame_saving)
• Discovery functions (discover_camera_ids, check_cti_file, add_cti_file)

Binding Class Patterns

When implementing camera support in a binding class, follow these patterns:

Basic Structure

class VideoSystems:
    """Manages video acquisition from cameras.

    Args:
        data_logger: DataLogger instance for timestamp logging.
        output_directory: Directory path for video file output.
        camera_configuration: Camera settings from system configuration.

    Attributes:
        _camera: VideoSystem instance for frame acquisition.
        _camera_started: Tracks whether acquisition has started.
    """

    def __init__(
        self,
        data_logger: DataLogger,
        output_directory: Path,
        camera_configuration: CameraConfig,
    ) -> None:
        self._camera: VideoSystem = VideoSystem(
            system_id=np.uint8(51),
            data_logger=data_logger,
            output_directory=output_directory,
            camera_index=camera_configuration.camera_index,
            camera_interface=CameraInterfaces.HARVESTERS,
            # ... other parameters from configuration
        )
        self._camera_started: bool = False

    def start(self) -> None:
        """Starts frame acquisition (does not save frames)."""
        if self._camera_started:
            return
        self._camera.start()
        self._camera_started = True

    def start_saving(self) -> None:
        """Begins saving frames to disk."""
        self._camera.start_frame_saving()

    def stop(self) -> None:
        """Stops acquisition and releases resources."""
        if self._camera_started:
            self._camera.stop_frame_saving()
        self._camera.stop()
        self._camera_started = False

Key Patterns

_started

__del__

stop()

System ID Allocation

Current allocations in mesoscope-vr:

51

62

Configuration Requirements

Camera configuration must be defined in sl-shared-assets before implementation.

Required Configuration Fields

*_camera_index

int

*_camera_quantization

int

*_camera_preset

int

Configuration Dataclass Pattern

@dataclass()
class SystemCameras:
    """Camera configuration for the acquisition system."""

    camera_index: int = 0
    """Camera index from the discovery function output."""

    camera_quantization: int = 15
    """Quantization parameter (0-51). Lower values produce higher quality."""

    camera_preset: int = 5
    """Encoding speed preset (1-7). Maps to EncoderSpeedPresets enum (SLOW=5 default)."""

Troubleshooting

Camera Not Detected

1. Verify driver software is installed
2. For Harvesters: check CTI file is configured using

   get_cti_status()

   MCP tool
3. Check physical connections and power
4. Run

   list_cameras()

   MCP tool to see available cameras

Encoding Failures

1. Verify FFMPEG installation using

   check_runtime_requirements()

   MCP tool
2. Check GPU availability for hardware encoding
3. Monitor GPU memory and thermal status

Frame Drops

1. Reduce

   display_frame_rate

   or disable preview (

   None

   )
2. Use faster

   encoder_speed_preset

3. Increase

   quantization_parameter

   (reduces quality)

Process Crashes

1. Ensure DataLogger is initialized before VideoSystem
2. Verify output directory exists and is writable
3. Check available disk space

Implementation Checklist

Before integrating cameras into an acquisition system:

- [ ] Verified ataraxis-video-system version matches requirements
- [ ] Confirmed FFMPEG and GPU availability using check_runtime_requirements() MCP tool
- [ ] Configured CTI file (for Harvesters cameras) using get_cti_status() and set_cti_file() MCP tools
- [ ] Verified cameras are detected using list_cameras() MCP tool
- [ ] Recorded camera indices from discovery output
- [ ] Tested camera with interactive session using MCP video session tools
- [ ] Created configuration dataclass in sl-shared-assets
- [ ] Implemented binding class with lifecycle methods
- [ ] Allocated unique system IDs for each camera