GERTIE

Grouped Electronic Real-time Imaging Environment

An 8-camera distributed Raspberry Pi system for high-throughput specimen digitisation, developed at the Natural History Museum London.

GERTIE captures synchronised 12.3 MP images from 8 networked cameras simultaneously, providing real-time video preview and centralised control through a Qt GUI. Built as a 3,000 GBP open-source alternative to 30,000 GBP commercial imaging systems.

---

🚀 Quick Start

New to GERTIE? Start here:

1. INSTALL.md — Complete deployment guide (USB setup, control1 configuration, node deployment)
2. Project Structure — How the codebase is organised
3. Keyboard shortcuts — Control the GUI

For development:

• See Project Structure below
• Check CHANGELOG.md for recent changes
• Review GitHub Issues for known issues

License: MIT (see LICENSE)

---

📚 Table of Contents

• Quick Start
• System Architecture
• Features
• Tech Stack
• Performance
• Deployment
• Network Requirements
• Project Status
• Keyboard Shortcuts
• Project Structure
• Background
• Contributing
• License

---

System Architecture

                    +------------------+
                    |    control1      |
                    |  (Qt GUI host)   |
                    |  + rep8 (local)  |
                    +--------+---------+
                             |
                     Gigabit PoE Switch
                             |
        +------+------+------+------+------+------+
        |      |      |      |      |      |      |
      rep1   rep2   rep3   rep4   rep5   rep6   rep7
      IMX477 IMX477 IMX477 IMX477 IMX477 IMX477 IMX477

Each remote node (rep1-rep7) runs a Python camera service communicating with control1 over TCP. Rep8 is a local camera on the control node. The entire cluster operates air-gapped with no internet connectivity.

Features

• Real-time 8-stream video at 16+ FPS in grid view, 30 FPS in exclusive single-camera view
• Synchronised capture across all cameras with sub-second latency
• Per-camera controls: exposure, white balance, saturation, crop, with AE/WB lock indicators
• Guard-Then-Act safety: 2-click confirmation for destructive operations (AE/WB clear)
• Gallery with automatic refresh after batch capture
• Context-sensitive shortcuts bar adapting to grid vs exclusive mode
• Per-tile overflow menus for camera restart/reboot
• DNG raw capture support
• Automated file naming: device ID + timestamp + structured folder hierarchy
• Museum workflow compatible: BardecodeFiler and EMu ingestion support

Tech Stack

Component	Technology
GUI	Python, Qt/PySide6
Networking	Custom TCP controller/node protocol
Camera API	libcamera / PiCamera2
Sensors	IMX477 (HQ Camera), IMX296 (Global Shutter)
Services	systemd units with auto-restart
Hardware	Raspberry Pi 4/5, Gigabit PoE
Deployment	rsync + Bash verification scripts
CAD	Autodesk Fusion 360, FDM 3D printing

Performance

• System overhead reduced from 40% to under 10% through async streaming optimisation
• 8 concurrent camera streams at 16+ FPS (grid), 30 FPS (exclusive)
• 12.3 MP capture (4056x3040) per camera, full sensor utilisation
• Camera footprint reduced to 15 cm through custom 3D-printed mounts
• User testing: 5 testers, 255 specimens, 3,418 files, 22.51 GB

Deployment

GERTIE operates air-gapped. Deployment follows a USB transfer pipeline:

MacBook (development) --> USB drive --> control1 (production)
                                    --> sync_to_nodes.sh --> rep1-rep7

Pre-deployment verification: bash verify.sh (5 automated checks).

Network Requirements

IMPORTANT: GERTIE uses UDP-based messaging for video streaming and commands. This design is suitable for local network (LAN) deployment only.

Suitable Deployments

• ✅ Museum/lab networks (controlled environment)
• ✅ Single subnet, <100m distance
• ✅ Dedicated Ethernet (not shared with heavy traffic)
• ✅ Packet loss <1%

Not Suitable For

• ❌ Wide Area Network (WAN) or internet-facing
• ❌ Wireless networks (unless <50m line-of-sight)
• ❌ Congested shared networks
• ❌ Cloud-hosted controllers

Specifications

Minimum: 100 Mbps Ethernet switch, latency <50ms, CAT5e+ cabling
Recommended: 1 Gbps switch, latency <10ms, CAT6+ cabling

For complete messaging protocol details, see MESSAGING_PROTOCOL.md.

Project Status

• Phase A: 10/10 complete (v2 feature ports)
• Phase B: 5/7 complete (interaction refinements)
• Under consideration for national deployment across partner institutions

Keyboard Shortcuts

Key	Action
Space / C	Capture All
1-8	Toggle exclusive view
Esc	Grid view (all cameras)
R	Restart camera
S	Settings
G	Gallery
D	DNG toggle
F	Focus mode
Q	Quit

Project Structure

GERTIE_Qt/
├── src/              # Qt GUI application (PySide6)
├── node/             # Remote camera node code (synced to rep1-rep7)
├── shared/           # Shared configuration and utilities
├── local_camera_node.py  # Local rep8 camera handler (systemd service)
├── tools/            # Utilities (deployment, testing, etc.)
├── docs/             # Architecture and reference documentation
└── INSTALL.md        # Complete deployment guide

Background

GERTIE was built from scratch by a single developer with no prior programming background, progressing from zero Python experience to a deployed distributed system in 18 months. The project encompasses hardware design (Fusion 360 + 3D printing), software development (Python/Qt), network protocol design (TCP), service management (systemd), and deployment automation (Bash/rsync).

Contributing

Contributions welcome. Please:

1. Review the Project Structure to understand the codebase
2. Check GitHub Issues for known issues
3. Read CONTRIBUTING.md for development guidelines
4. Submit a pull request with clear commit messages

For questions about deployment or architecture, start with INSTALL.md or open an issue.

License

MIT License — See LICENSE for full text.

Copyright (c) 2024-2026 Natural History Museum London

GERTIE is provided as-is for research, educational, and commercial use. Please cite this work if you use it:

Crane, A. (2026). GERTIE: Multi-camera specimen digitisation system.
Natural History Museum London. Available at:
https://github.com/andrc2/GERTIE-Qt