motion capture, post-capture thoughts
  • I see this old (2015) thread, which is relevant: https://www.bensoftware.com/forum/discussion/1063/recording-constantly-stops/p1

    It would be nice that if SecuritySpy is already in the middle of a motion capture and there's a pause, but then motion starts again wile the post-capture timer is running, then the post-capture timer should be reset. This would neatly kill the cases where you have someone walk into frame, stop, then continue to move and the video stops halfway through the motion event.

    Really, it seems that post-capture time should mean "after the last motion event, continue recording 'x' seconds", not "after the FIRST motion event, continue recording 'x' seconds". This is especially true when a second recording will not begin because the post-capture time is still running.
  • Just an addendum to try to explain more clearly. This is an abbreviated event log:

    20200121143011 4 3 TRIGGER_M 256
    20200121143014 10 3 TRIGGER_M 256
    20200121143020 21 3 TRIGGER_M 128
    20200121143030 50 3 TRIGGER_M 128
    20200121143049 106 3 TRIGGER_M 128
    20200121143056 113 3 TRIGGER_M 128

    in my opinion, the post-capture timer should be reset after each and every TRIGGER event for the camera. If, for example, my pre-trigger time is 10s and my post-trigger time is 30s, then the resulting video file should start at 14:30:11 - 10s or 14:30:01, and then the last frame in the recording should be for 14:30:56 + 30s, or 14:31:26.

    As it is right now, it appears that the post-capture timer starts from the first trigger, making almost all of my videos exactly 40s long. And if there's a gap in the motion, then the video stops while there's still motion going on in the frame.

    I know it's not quite this cut-and-dry, because while MOST of my motion capture videos are 40s long, I do occasionally see ones as long as a minute or even a minute and a half.
  • The post-capture time operates exactly as you are describing that it should: the post-capture timer resets every time there is another trigger while it's already recording. This is why you are getting movie files that are more than 40s long sometimes.

    If you compare the triggers shown in the above event log, and the length of the movie files actually recorded during time time, what do you see? As you correctly state, with your settings (10 pre and 30 post), if the first trigger occurred at 14:30:11, and there were subsequent triggers less than 30s apart, with the last one occurring at 14:30:56, the result should be a movie file that starts at 14:30:01 and is 1m25s in duration.
  • Hm, perhaps there needs to be a "hysteresis" to the trigger classifiers. Something like "trigger level to *start* motion capture is 80%, but then the trigger level to keep the motion capture moving is only 40%".

    I'm trying to figure out a better way to log this so that I can get better data to you. Right now if a person drives into frame (and causes a trigger), then gets out of the car and moves around the vehicle (delivering a paper, whatever), and gets back in and drives off I have about an 80% chance that the motion capture video stops in the middle of this action, without a subsequent motion capture video of the rest of the action.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!