Camstack

Camera management stuff. Including keywords and TCP streaming from the stream source.

Gets installed with pip install -e .

Look at the scripts/startOCAM to get an idea. This creates the python control shell in ocam_ctrl.

Also includes the camera viewers - and their backups

Camera acquisition stack

Layout

./scripts folder: bash launcher scripts. Cleanup of the <cam>_ctrl tmux, and eventually launch the python camstack.cam_mains.<cam> main script.

./camstack/cam_mains folder: camera launcher python scripts.

• Define auxiliary tasks:
  • a tmux session
  • a command line
  • a cpuset and a real-time priority
  • and a sequencing order in which to be started/killed relative to other aux tasks.
• Instantiate the camera. A camera class is:
  • a subclass of the BaseCamera
  • with a specialized backend (USB, framegrabber, underlying API...)
  • with model specific definitions (C-Red One, C-Red 2, Orca, etc...)
  • with camera specific definitions (special cropmodes for Palila, GLINT, etc...)

Camera startup

Instantiating the Camera class eventually results in the camera freerunning in a shared memory and all the auxiliary tasks being started.

The camera launcher camstack.cam_mains.<cam> script, which runs in the <cam>_ctrl tmux session, eventually drops to an interactive python prompt. This is where you control the camera. A clean quit is performed by issuing the close() command.

Init sequence:

This is the general outline of what happens during the Camera Class constructor.

• kill_taker_and_dependents(): Allocate and clear all tmux sessions for framegrabber and auxiliary tasks
• init_framegrab_backend(): [BACKEND SPECIFIC] initialize resources for acquisition on the receiving end. This can be opening a serial port, configuring the acquisition size, etc...
• prepare_camera_for_size(): [BACKEND AND CAMERA SPECIFIC] set the camera crop mode. Performs tasks that need the camera control channel open but that need to be done before the acquisition starts.
• _start_taker_no_dependents():
  • _prepare_backend_cmdline(): [BACKEND SPECIFIC] prep the shell line to be run in the <cam>_fgrab tmux. This is a minimal chunk of C to make the camera freerun to a SHM.
  • Start said cmdline and begin acquisition.
  • Adjust real-time priority and cpuset
• grab_shm_fill_keywords(): [CAMERA SPECIFIC] get a python handle to the freshly created SHM (by the framegrabbing process), and proceed to populate FITS keywords specific to the camera. They'll propagate through TCP all the way to the logger. This is not backend specific, the access is done through pyMilk, but the exact keywords are camera specific.
• prepare_camera_finalize(): [CAMERA, BACKEND SPECIFIC] Finish configuring the camera for the acquisition mode you want, with those last commands having to / allowed to be issued after the camera freeruns. Such as setting fps, integration time, NDR, exttrig for some models.

See a more extensive example in this page

Changing camera "mode"

Changing "mode" really means changing crop size. The framegrabber has to be reconfigured, all SHMs re-instantiated with their new size, etc. Pretty much all steps above are called in the same order. This is done without quitting at the <cam>_ctrl command prompt, by calling set_camera_mode(some_predefined_mode_id).

For dumb cameras (acquisition channel but no control channel), the FG acquisition can be set to an arbitrary size dynamically by calling set_camera_size(height, width).

Recap

Class tree:

• BaseCamera
  • EDTCamera
  • CRED1
    • Apapane
  • CRED2
    • Palila
    • GLINT
    • Kiwikiu
  • OCAM2K
  • NUVU
    • Kalao
  • AutoDumbEDTCamera
    • AutoAndor897
    • First
    • Vampires
  • DCAMCamera
  • OrcaQuest
    • FIRSTOrcam
    • AlalaOrcam
  • SpinnakerUSBCamera
  • BlackFlyS
  • FLIR_U3_Camera (broken, use FlyCaptureUSBCamera::(Grasshopper3|Flea3) instead)
  • FlyCaptureUSBCamera
  • Grasshopper3
  • Flea3
    • VampiresPupilFlea
    • FirstPupilFlea