Motion detection
Motion sources, server-side motion, zones, and the motion grid.
Source
A camera can report motion in several ways:
- Server-side motion: the node compares video frames inside motion zones or the motion grid.
- Camera events: the camera's own ONVIF motion events.
- Vendor integration: for example Dahua / Amcrest smart events.
- AI detection: a detected moving object.
The camera's Motion source chooses which one of these decides what Motion mode keeps, so the timeline doesn't get duplicate entries for the same event. Auto picks the richest one that is set up. Event tag rules and AI object labels always tag the timeline regardless of this choice.
Server-side motion
Server-side motion on the camera's page turns the node's own pixel motion detection on or off. Keep it on to tag plain motion on the timeline as a fallback, even when another source gates recording. Turn it off to save the CPU it uses.
Zones
A camera's Zones page has two methods. Only one is active at a time, and switching keeps the other's settings.
Grid: click cells to mask them out (red). Cells turn amber while there's motion in them, which helps you find a swaying branch or a flickering light to mask. Changing the grid size clears the mask.
Polygon zones:
- Motion: an area to watch.
- Ignore: an area to exclude (a tree, a public sidewalk). Pixels in an Ignore zone never count, even where it overlaps a Motion zone.
- Privacy: an area to black out. Draw these under 🔒 Privacy masks, which works whichever motion method (Grid or Polygon zones) the camera uses. It's hidden in recordings, live view (both streams), snapshots and exports. Events still happen behind it as usual (their snapshots are masked like the video) unless you turn on Also ignore detections inside privacy masks for that camera, in the Privacy masks view. The mask follows the shape you draw, with a small margin so no edge shows through. A camera with a privacy zone has its video re-encoded while recording, which uses some CPU (or the GPU, when the node has a supported one).
- Camera-side motion zones can be saved but are not active yet.
To reshape a zone, select it in the list and drag its corners, then Save. While drawing, right-click removes the last point. Hovering a zone in the list highlights it on the picture.
Sensitivity is the percentage of the zone's pixels that must change. Lower is more sensitive. Wide zones need lower values, because real movement covers only a small part of them. Start low (around 3%) and raise it only if wind or lighting changes trigger it.
Also in the app under Help. View source on GitHub.