simple-cam_libcamera

A dual-camera video capture and display application for the Raspberry Pi, built on libcamera for capture and OpenCV for frame processing and display. It captures two camera streams simultaneously, converts each frame from YUV420 to BGR, and shows them side by side in a single window.

The project grew out of the upstream libcamera simple-cam example and was extended to drive two cameras at once. Alongside the main application, the repo contains a set of small C++ learning sandboxes (Qt signals/slots, inheritance, threads, OpenCV parallel_for_) used while building it.

Status: work-in-progress / learning project. It runs and shows both feeds, but color correctness and per-frame performance are still being tuned. See Known issues.

---

Features

• Enumerates and acquires two cameras via libcamera's CameraManager.
• Configures both as 1280×720 YUV420 viewfinder streams with auto-exposure.
• Maps capture buffers into CPU memory (mmap via the Image helper).
• Manual YUV420 → BGR conversion with OpenCV (plane split, chroma upsample, merge, cvtColor).
• Concatenates both cameras into one window once a frame from each is ready.
• libevent-based event loop with a capture timeout.

Repository layout

Path	Description
dual-camera.cpp	Main program — captures two cameras and displays them side by side.
simple-cam.cpp	The original single-camera example (built as single-camera).
image.h / image.cpp	mmap-based wrapper for accessing frame pixel data (from upstream libcamera).
event_loop.h / event_loop.cpp	libevent-based event loop (from upstream libcamera).
CMakeLists.txt	Build configuration (the single canonical build system).
buildLibcamera.sh	Installs dependencies and builds/installs libcamera from source.
projectNotes.md	Author's study notes (YUV420 layout, pointer math, lambdas, CMA, profiling).
parallel_for_example/	OpenCV parallel_for_ + TBB experiment. See docs/examples.md.
SignalAndSlotExample/	Qt signals/slots study. See docs/examples.md.
InheritanceExample/	C++ inheritance study. See docs/examples.md.
ThreadFolder/	std::thread worker study. See docs/examples.md.
Screenshots/	Reference screenshots used by the docs.

Documentation

• Building & installing — dependencies, building libcamera, and building this project.
• The dual-camera application — how it works, how to run it, known issues.
• Profiling — gprof and perf workflows.
• Learning examples — the four standalone study sub-projects.

Quick start

This project targets a Raspberry Pi running Raspberry Pi OS with two cameras attached.

# 1. Install deps and build/install libcamera (first time only)
./buildLibcamera.sh

# 2. Build this project (also installs nothing system-wide)
cmake -B build
cmake --build build

# 3. Run the main dual-camera app
./build/dual-camera

See docs/building.md for prerequisites (OpenCV, libevent) and details, and docs/dual-camera.md for runtime requirements (e.g. the Wayland/X11 display note).

License

The files inherited from upstream libcamera (simple-cam.cpp, image.*, event_loop.*) carry their original SPDX headers (GPL-2.0-or-later / LGPL-2.1-or-later, © Ideas on Board Oy / Google Inc.). dual-camera.cpp is derived from simple-cam.cpp and inherits GPL-2.0-or-later.