I understand that each none added to the pipeline also requires processing and it's desiderable to have as few nodes as possible (please correct me if I'm wrong).
Getting a recording out of the pipeline can be achieved either by using the VideoEncoder node or by writing the actual frames to a video file using cv2.VideoWriter
Could someone explain me what advantages and disadvantages would be for each way.

  • erik replied to this.

    Hi bogdan ,
    That's correct. Note that video enocding is done on it's own hardware asic, so it wouldn't take too much of other resources except ram.
    Difference is that if you run it on OAK, the whole compute is on the OAK, as opposed to using resources on the RPI. I would also guess that RPI isn't powerful enough to do eg. 4k 30FPS, even if not using the RPI for any other compute.
    Thoughts?
    Thanks, Erik