-
-
Notifications
You must be signed in to change notification settings - Fork 193
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Some ideas not sure if there are already reported #855
Comments
There is some fresh documentation on this in the
Good suggestion, i have the same problem with my cameras.
Viseron v3 is currently in beta, which includes 24/7 recordings and with it a timeline view. Which version of the Docker container are you running? |
Thanks will try that.
I was thinking like something setting resolution for the live camera feed. Choosing for every camera a resolution so every one of them looks similar on the live feed. But yes also if the aspect ratio of the live feed resolution (after you changed it) is the same as the main stream them the still image on the recordings will also match the video and will make it look better when someone tries to access the recordings.
I am using the latest tag which in docker says is 15 months old. I will try to switch to the tag you mention above and let you know of the results. |
Hi @roflcoopter , I have just update to version 3.0.0b12. I like the new interface and I was able to create a user where I can access my cameras now with more security. Some things I notice after the upgrade:
Update: 13/12/24 Hi @roflcoopter , I had to revert to the previous stable version for now for the following reasons:
Let me know if you need me to test anything on my system to pinpoint where the above "issues" are. |
Hello, @phanosp and @roflcoopter,
I can confirm, that with build 3.0.0b12 this is happening every single time. first 10-20 seconds of event are lost. As I understood from the new Events feature description by @roflcoopter, Viseron right now saves events not by whole recordings, but by 10 seconds (or something close to that) recordings and takes them back from BE with playlists. I assume, that ether those first recordings are failed to save, but entries in db are made, ether they are deleted by some kind of cleanup process falsely. In the latest build playback are acting differently: Playback progress starts from first second, but its clear that its missing first (or maybe two) batches of recording. Usually thumbnail image (before triggering playback) shows actually good moment (moving vehicle or something), but by playing the recording looks like false positive and nothing important recorded. Additionally latest build looks like has also another issue, where last 15 seconds are also missed to record, because progress bar shows more time, but footage is stuck. @roflcoopter If you would like to get logs, do some debugging, let me know. A have to say, that @roflcoopter you are doing great job here, impressive progress. Keep going, if you need help with testing latest dev, let me know, ready to do that (I am actually doing that last few weeks 😄 ). |
Hi @kilodelta1 , basically what you described is what I experience when trying out the build 3.0.0b12. @roflcoopter if you need logs to fix any bugs lets us know. |
Yeah old reordings are not migrated.
Nothing has changed in this regard, so it must me a coincidence. What are you using to trigger recordings?
Do you mean its slower to load or does it stress your host more?
Will look into the recordings starting too late. I have no issues with that but ill see what i can find. Recordings are stored in Note that the documentation does not reflect the current state, rather a future state |
There might be some confusion here what actually ends up as a recording vs just snapshots. I have had some good discussions in #837 as well trying to clarify some things. All in all, the current state of the beta might not be very intuitive for new users but i intend to work on that before the final release so i appreciate your feedback. I did some testing and here you can see the timings of a detected object event: As you can see there is only 2 seconds that differs between the detection and the timestamp that is seeked to when i click it. Do you think you could show me by either screen shot or a short screen capture so i can get a better understanding of what the issue is? |
Hi @roflcoopter
I am using mog2 like the following:
Maybe the issue was misinterpreted though. After reading what @kilodelta1 wrote maybe the recording is the same but is not being played back correctly, it starts after some seconds and finishes without playing all the recording, which might result in the video not showing the actual trigger event.
Both. I see the interface loading slower and the cpu going higher. Ofcourse I am using an old cpu but still the old stable version works ok on my system. Does the new version requires a newer cpu with new requirements?
I am only using mog2 as shown above. In the current version it works fine but maybe in the new version I need to use something more? So an Event recording (https://dev--viseron.netlify.app/docs/documentation/configuration/recordings) is not something that can be trigger only by motion but it needs to have an object detector? Am I understanding correctly? If this is the case then in the new version is there a way to do just motion detention while using the new version? What I experience in the new version is that although that an event is shown in thumbnail (say a car passes in front of the camera), when I try to play the recording it starts around 10-20 seconds after the initial frame and sometimes stops even before the end. The car is not shown at all in the video although is shown in thumbnail. Unfortunately I have no video to show right now because when I notice the issue I could not find the location of the recordings of the new version to save them, so I deleted everything and make a fresh install of the stable version so I get a working system. If you wish me to test again I can do so so please let me know. |
The newer version does require some more juice since it is using some heavy DB queries.
No you can still use only a motion detector, no obejct detector is required.
Could you maybe setup a second Viseron instance running the v3 beta to do some testing? |
Yes I can do that. Do you want me to run it and get the logs and sent them to you or you want specific thinks to test before sending the logs? |
Would be great to have the full logs from startup up until the end of a recording. Would be greate if you also could show the behaviour you described in the web interface either with a screen capture or screenshots |
Hi @roflcoopter, I have just set up the version 3.0.0b12 and will run it for some time and will report back with the logs and perhaps some videos. |
Hi @roflcoopter , just a little update. In the following there are the logs since I start up the docker viseron and a video showing the problem I mentioned earlier. Although the screenshot on the video clearly shows a vehicle was present during the even (up side of the video) during the playback there is no vehicle. Also if you notice the video jump starts around 9 seconds into the video. Perhaps this is why the vehicle is not being shown here? Video_2025-01-11_14-08-15.mp4 |
Sorry for the late reply. Thank you so much for the video and the logs. Will need some time to think on what could cause this. One thing i noticed is that the datetime in the video is way off from the recordings start time. Is it actually out of sync or is your timezone on your client set to something different than the cameras? |
What machine are you running Viseron on btw? |
Hi @roflcoopter, the datetime of the cameras are wrong. I tried many times fixing it along with setting the correct timezone but every time a camera is rebooted it is being reset to its default time. The rest of the settings are fine thought after reboot. Not sure what could cause this but for now I ignore it. I simply run it and view the datetime that viseron reports that is correct. I am running the viseron on an old machine with an intel atom D520 and 2 GB of ram with debian 12 os (no gui). For version 2.xx the cpu is around 70-80% but works just fine. Nothing else is being run on this machine. For version 3.0.0b12 the cpu is arround 80-90%. Maybe there is some hardware requirement that the new version has and my old cpu does not have? |
In v3 a lot more is happening in the background so yeah it will use more resources than v2. It seems that your machine is too slow to process the produces segments which is causing other issues down the line. My guess is that your RAM is too low which causes the system to start swapping. Not much i can do about that unfortunately |
Thanks @roflcoopter for the info. I will be building another machine soon with a xeon cpu and 64GB of ram (most parts already purchased) but I expect to assemble it the latest by the end of March. Until then I can always check again on another machine that is currently not being used, an i5-3570k cpu with 16GB of ram and sent you the logs if the problem persist. Do you think is worth it or you are sure that the problem described above by me and @kilodelta1 is caused by a weak cpu and very little ram? |
Nice, that should be plenty!
If you have the time, that would be great. I am pretty sure your issues are caused by a weak machine, i run the beta on my server with i3-9350K and 64gb of ram and i do not have any issues similar to yours. |
ok @roflcoopter I will test as soon as I go home and get back at you. I am currently abroad far away from Europe and will return later this month. I will test then and if I notice the problem I will get back to you with the logs. |
Could you do what @phanosp did above, make a screen capture and send over the logs? You can enable debug logs like this: logger:
default_level: debug |
The text was updated successfully, but these errors were encountered: