Four IP cameras, one of them acting strange

[BiloxiGeek]

I have four Grandstream GXV-3672-FHD cameras, two of them are the base model, two have a 3.6mm lens for a wide angle view. One of the wide angle cams is throwing a lot of events due to smearing. The other 3 don't do this at all, or at most very very infrequently.

For instance, I cleared out all the events about 30 minutes ago, 3 cameras have detected no events in that time, the one odd-ball is showing 8 events. When I look at the alarm frames they're all due to smearing in the image.

I've checked settings, all the cameras are configured the same, all the monitor definitions in zoneminder are the same. Can't figure out why this one camera is doing this.

Any thoughts on what could cause this? Could it just be bad hardware?

Added: The attached images are from the low def stream that I set up as modect, the hidef stream is set up nodect and linked to the lowdef monitor. When I look at the frames from the same event on the hidef side I don't see smearing in those frames. I've considered just dumping the lowdef streams and do everything in hidef but I'd be concerned about the extra workload on the system having to analyze 1920x1088 frames instead of the 320x176 frames.

[linuxsense]

Are you using ffmpeg? I get the same issue with one of my IP cams but its fine if I use libvlc.

[linuxsense]

This might be of interest to you: http://www.zoneminder.com/forums/viewto ... 30&t=21911

[BiloxiGeek]

Yep, using ffmpeg, I tried libvlc but it was very unstable for me. My monitor definitions all use the "?tcp" in their rtsp URL.

I submitted a ticket about this to Grandstream, they suggested a factory reset. I did the reset, got the camera back to the same configuration it was in before and that my other two identical model cameras are set for, and still seeing smearing on just this one. It's really starting to look like a hardware issue in the camera itself.

[PacoLM]

How long is the cable you are using for the camera that smears? I suggest to check the cable and the category (use a CAT5 minimum).

[linuxsense]

If you ping the cam for 5 minutes or so do you get any packet loss?

[BiloxiGeek]

The cable is actually the shortest of the four cameras that I've got hooked up. The longest is only about 80 feet. It's good Cat5e cable, I terminated both ends myself. Been building Cat5 cables for just about 15 years. I got lots of experience there. There's three other cameras on the same switch, none of them are doing the smeared images, they're all similar models and all configured the same.

Code: Select all

--- icecam5 ping statistics ---
122 packets transmitted, 122 received, 0% packet loss, time 121065ms
rtt min/avg/max/mdev = 0.198/0.333/2.204/0.241 ms

[linuxsense]

Does the oddball camera give the same results if its the only one feeding zoneminder?

A few weeks back I was getting some funky images from one of my IP cams if the switch it was connected to had a fair amount of traffic on it. Switch seemed fine but I had a few spares so I swapped it out...problems went away. The one that I replaced was a pretty decent gigabit unit but it seemed to fall apart if one of its ports had large amounts of traffic on it. I am sure its unlikely that you are running into a similar situation, just tossing the idea out there in case you start running out of things to troubleshoot.

[BiloxiGeek]

Ran the ping a bit longer this time:

Code: Select all

--- icecam5 ping statistics ---
610 packets transmitted, 610 received, 0% packet loss, time 609350ms
rtt min/avg/max/mdev = 0.215/0.325/13.485/0.553 ms

Did see something interesting during that ping.

Code: Select all

64 bytes from icecam5 (192.168.1.235): icmp_seq=155 ttl=64 time=0.273 ms
64 bytes from icecam5 (192.168.1.235): icmp_seq=156 ttl=64 time=13.4 ms
64 bytes from icecam5 (192.168.1.235): icmp_seq=157 ttl=64 time=0.315 ms

and

Code: Select all

64 bytes from icecam5 (192.168.1.235): icmp_seq=222 ttl=64 time=0.278 ms
64 bytes from icecam5 (192.168.1.235): icmp_seq=223 ttl=64 time=2.63 ms
64 bytes from icecam5 (192.168.1.235): icmp_seq=224 ttl=64 time=0.275 ms

Got a smeared image which triggered an event when those happened, didn't lose the pings, but they were awfully slow.

[BiloxiGeek]

Does the oddball camera give the same results if its the only one feeding zoneminder?

Just disconnected the other cameras from the POE switch. Will have to let it run for awhile to see if that makes a difference.

[linuxsense]

Interesting to see the latency spike like that.

What framerate are you running on the camera? I have found that limiting the framerate on the camera and leaving the ZM fps options empty to be much more reliable. With one of my IP cams setting the fps on the ZM side causes nothing but problems.

[BiloxiGeek]

All the cameras are set (within the camera setup) to 10fps, I've matched that on the zoneminder side. (When I shouldn't have)

Or at least I thought I had. I found the oddball camera set to 10fps along with another camera. I must have gotten used to setting 10fps whenever I saw the option. I've reset it to be blank on the zm side, will let it run for awhile and see what happens.

[linuxsense]

That might help out quite a bit. I had a cam at 10 fps and with ZM also set to 10 fps it would hang the stream and provide funky images. Leaving the ZM fps setting blank sorted it out. I think that although set to 10 fps on the camera it was actually running 10.4 fps (give or take) so ZM was trying to limit it.

[BiloxiGeek]

That might help out quite a bit. I had a cam at 10 fps and with ZM also set to 10 fps it would hang the stream and provide funky images. Leaving the ZM fps setting blank sorted it out. I think that although set to 10 fps on the camera it was actually running 10.4 fps (give or take) so ZM was trying to limit it.

Yep, when I look at a montage of low def streams they are all showing up as running at 10.27 or 10.28fps. It's looking a whole lot like that was the borked up bit. Fifteen minutes later and no false events, no smearing that I've been able to see.

Amazing how one little slip of the keyboard will drive you frakkin nuts. On the up side, I now have 5 of my six cameras set up in ZM, and showing a load average of about 1.5 on a Core i7 3.4GHz, 8 core, 16GB system. I'm liking this IP camera set up more and more each day.

[Biggus]

Is the firmware the same across all cameras?