Cameras

MSW records video through a pluggable camera backend declared per setup under the cameras: block. Two backends ship:

• rce - Raspberry Pi camera ensemble (freely moving rigs): one or more networked RPis each running an RCE agent, coordinated by a conductor. Provided by rpi-camera-ensemble.
• flir_bonsai - FLIR/Spinnaker cameras driven by Bonsai subprocesses (head-fixed rigs, Windows acquisition machines). Provided by msw-flir-bonsai.

cameras: null (the default) means no cameras are configured and the task runs without video.

The full key reference for both backends lives in Setup Config → Camera config. This page is the orientation and per-backend bring-up checklist.

Choosing a backend

	rce	flir_bonsai
Typical rig	Freely moving	Head-fixed
Hardware	Networked RPis + ribbon cameras	FLIR/Spinnaker cameras on the acquisition PC
Sync	TTL barcodes per agent	TTL barcodes; per-camera serial sidecar
Driver	RCE agent + conductor	Bonsai workflow subprocess per camera

RCE bring-up (freely moving)

1. Flash each RPi and install the RCE agent package; bring each Pi onto the rig network with a known, reserved address (reserved in your lab network config).
2. Author the ensemble YAML describing each agent (host, camera id, role) and place it under msw_configs/device_configs/cameras/.
3. Point the setup at it:

   cameras:
     backend: rce
     config: device_configs/cameras/<setup>.cameras.yaml
   

   config is absolute or relative to msw_configs/.
4. Verify GPIO in/out, the agent, and the conductor on each Pi before a session.

Record the physical mapping (camera position ↔ camera id ↔ Pi ↔ network address) for your rigs in your lab's private config repo, not here - it is deployment-specific and not part of the published package.

FLIR + Bonsai bring-up (head-fixed)

1. Install Bonsai and the Spinnaker/FlyCapture SDK on the acquisition PC.
2. Resolve the Bonsai executable: msw flir find-bonsai (or set the BONSAI_EXE environment variable).
3. List cameras to resolve enumeration indices: msw flir list-cameras.
4. Configure the setup:

   cameras:
     backend: flir_bonsai
     bonsai_exe: C:\Users\<user>\AppData\Local\Bonsai\Bonsai.exe
     cameras:
       - index: 0
   

   One Bonsai subprocess is launched per camera entry. Per-camera optical parameters (gain, shutter, exposure) are set inside the Bonsai workflow XML, not in this config.
5. Smoke-test: msw flir test-record, then run a task with the backend set.

Enumeration index is not guaranteed stable across reboots; the per-camera serial is recorded in the FLIR session sidecar (see Setup Config → FLIR session metadata sidecar).

Synchronisation

Both backends rely on TTL barcodes for offline alignment of video against Bpod and ephys: the same 37-bit timestamp pulse train is distributed to each camera backend so frames can be matched to behavioural and electrophysiology streams after acquisition.