A real-time camera capture plugin developed in response to a live industry brief set by SimCentric - a UK based simulation software company - as part of a university wide faculty project. Four other companies set briefs alongside SimCentric:this was the chosen brief.

The brief centred on wildlife crime scene investigation, requiring tooling that could operate inside a live simulation environment, capturing camera frames at runtime, attaching structured metadata, and feeding that data into downstream detection or analysis workflows.

 

My Role

Core plugin development was implementing the real-time capture functionality, building the metadata logging system, and integrating the object detection workflows. Also responsible for integration testing across capture and metadata pipelines. The project was delivered by a team of approximately ten students, each with a defined role.

 

Technical Challenges & Solutions

Debugging runtime behaviour working inside a live simulation environment introduced a class of issues that don't surface in standard development - incorrect values being passed between systems, missing node connections in Blueprints, and variable naming errors that only became apparent at runtime. These weren't complex architectural failures, but in a real-time system they could cause silent data errors or unexpected behaviour that was difficult to trace.

The approach was systematic use of Unreal Engine's debug tooling: printing variable states to the viewport during runtime to verify values at each stage of the pipeline, and isolating individual systems to confirm correct behaviour before integration. This made it straightforward to identify exactly where a value was wrong or a connection was missing, and resolve it precisely rather than by trial and error.

No frame drops occurred during capture. The plugin was designed to operate without interrupting simulation execution, and this held throughout development and testing.

Metadata architecture The metadata system was built around Unreal Engine's Get functions, which identify and retrieve information about objects within the camera's view. This data is then structured and output as JSON. The architecture is intentionally flexible - metadata fields are not fixed, and can be expanded or tailored to the requirements of any given simulation environment. Whatever the camera sees, the system can be configured to capture and log it.

 

Key Features

• Real-time image capture from within a live Unreal Engine session

• Structured metadata logging - timestamps, coordinate data, and object identification

• JSON output with infinitely expandable and configurable metadata fields

• Object detection workflow integration via Unreal Engine Get functions

• Designed to operate without interrupting simulation execution

• Debugged and validated through runtime variable printing and engine debug tooling

 

Status

University project delivered and presented to SimCentric. Active development ongoing — current focus on improving detection accuracy and expanding metadata output fields.

​

Unreal Engine 5 Blueprints Real-Time Systems Metadata Logging Object Detection JSON GitHub Agile