Additional info: VMS is on a separate machine. so are the recorders. The hardware servers are Dual Xeon 2,4 GHz processors. SAS disks and plenty of RAM (16 on the recorders and the Base VMS machine). The hardware never seem to choke CPU is usually around 15% of utilization and RAM is 6 GB or so at most ... 2 Disk in RAID 1 for OS RAID 1 for Live retention and RAID 5 for archives. Works smoothly except for the "ghosting".
When you say ghosting, you mean people are blurred? Like so:
Or do you mean the recorded video starts with the person already inside the frame?
What is your recorded frame rate? If they are 'popping' in and out that could be a very low frame rate issue.
Video starts with person already in the frame so you have a view with a person (or several) just popping in the scene. They disappear also the same way . Frame rate is 10 fps. No one's running :) ... Jerkiness and eventually people popping in or out ...
IPVMU Certified | 11/12/14 08:35pm
What vesion of ONSSI are you running? The CS and lower code levels have a slightly different behaviore than the ES/LS versions.
In either case you will want the Pre recording active. The time defined should be enough so you can see the object coming into the frame.
Note that the PRE recording works by writing constantly to the storage subsystem and when the Motion event is detected the useful data is kept. If your storage system can not handle the extra IOs and space needed for the prebuffering then there will be performance consequences.
CS. I am interesed in recording at 1 fps most fo the time and bump it up to 15 fps once motion is sensed. I will check if the CS can do this.
I wan also thinking about the Pre recoding as well and will see how it affect total retention time...
Will give you my finding in the next 2 days
Keep suggestions coming please
IPVMU Certified | 11/13/14 01:55pm
The Total retention time 'available' will be affected by how much Pre and Post recording you set up...along with the instance counts of motion.
Be sure to properly 'tune' the motion detection settings for each and every camera so that you do not get more motion than what is needed.
Once tuned... monitor the database size for a couple of days to determine what total size your storage can accomodate.
Note that the storage Total Usable space is not the value to use in this calculation. Leave 20% of it available for the Archiving and maintenance functions.
I believe in CS you want to set the recording to "Always" 15 fps, then there was something like record key frames only, then on motion record all frames. Key Frames will give you your time lapse recording, so 1 -.33 IPS then on motion it will record your full frame rate. Then give yourself a few seconds of pre recording, and post recording that should smooth your video out.
just remember i think it takes a few minutes for it to take effect after you restart the services.
I believe the problem is solved thanks to you people! Pre-Recording and Post Recording helped. We also defined much more carefully the field within which the camera would analyze motion detection (eliminating ceilings, side walls no one except spiderman could walk on, etc...) .
I will ask OnSSi but I couldn't fid the setting on the IS that would allow 1 fps and with motion to go to 15 fps ...