Workaround: Linking dual-lens triggers (Fixed vs. PTZ)

[origin2000]

I’m currently setting up the TP-Link/Tapo C545D and I’m trying to solve a bit of a hardware-specific puzzle.

The C545D is a dual-lens camera:

Lens 1 (Fixed Wide): Fully supports ONVIF and motion events.

Lens 2 (PTZ): Tracks motion brilliantly in the Tapo app, but it’s 'silent' over ONVIF—it doesn’t expose its own events.

The camera's internal logic is already there—the fixed lens triggers the PTZ lens anyway. Since the PTZ RTSP stream is already working in lightNVR, I was wondering: Is there a way to use the ONVIF motion events from the Fixed Wide lens as a trigger for the PTZ recording?

Basically, I want to 'slave' the recording of the PTZ stream to whatever the fixed lens detects. I realize there’s a small downside: if a moving object leaves the Fixed Wide's field of view, it will send a 'no motion' signal and stop the recording, even if the PTZ lens is technically still tracking it.

But honestly? Having the PTZ record while the target is in the main frame would already be a huge win for me. Is this kind of cross-stream mapping something lightNVR can do (or might do in the future)?

Any feedback is welcome!"

[origin2000]

I checked the ONVIF Profile S datasheet for the C545D (onvif.org). It confirms the device uses a single RuleEngine/CellMotionDetector/Motion topic for the entire unit. Since the PTZ lens (Stream 6) is 'event-less' in the XML, would it be possible to trigger its recording based on the Motion event coming from the global device/fixed lens?

[matteius]

Check out: #336

Any feedback will be helpful and I don't have a setup like that.

[origin2000]

This is awesome! I checked #336 but I struggle where and how to put the config. Did comment there too.

[matteius]

Change pushed to main (in upcoming release) that exposed it in steams admin UI -- I missed that part.

[origin2000]

Matt, don't know if you want testing at #336 or here. Let me know, please. Here my testing results:

1. Motion Start & Propagation (Working):
   The Master (wide) correctly detects motion and immediately propagates the event to the Slave (ptz). Both recordings start synchronously.

[23:21:28] [Detection] [c545d_wide] ONVIF Detection: Motion detected for c545d_wide
[23:21:28] [Detection] [c545d_wide] Propagated motion event (start) from 'c545d_wide' to linked stream 'c545d_ptz'

2. Motion Stop Detection (Working):
   The Master also successfully detects the end of motion and sends the stop signal to the Slave.

[23:21:41] [Detection] [c545d_wide] ONVIF Detection: No motion detected for c545d_wide
[23:21:41] [Detection] [c545d_wide] Propagated motion event (end) from 'c545d_wide' to linked stream 'c545d_ptz'

3. The Issue: Slave ignores Stop signal:
   Despite the propagated "end" event, c545d_ptz fails to stop. It continues recording in 30-second chunks, while c545d_wide stops correctly.

[23:21:46] [Detection] [c545d_ptz] Ensuring detection-based recording is active for stream c545d_ptz
(Note: "Recording stopped" appears for wide, but is missing for ptz).

Logs show both wide and ptz performing independent PullMessages on the same ONVIF endpoint (/event-1026_1026).
Question: Does this dual-polling make sense? Does a PULL clear the camera's event buffer, so the two processes are competing for the same messages?