-
Semper Vaporo
-
- Posts: 331
- Joined: Sat Oct 09, 2010 11:51 am
by Semper Vaporo » Wed Feb 18, 2015 3:13 pm
WOW, nice specs for your system... I'm jealous!
I don't think that turning off saving of temporary video would do anything for you, even if you could. The passing of video data from the camera to the disk is trivial compared to the CPU work required for motion detection and "Person vs Object" determination to decide if it is not worth being saved even temporarily.
To free up CPU cycles, you will need to do one of two things.
1) Turn off cameras for certain hours of the day where motion detection is not useful. I realize that may not be realistic, depending on your particular circumstances. For example: if you have a camera with a view of your front porch and another with a view of the walkway to the porch, you might decide that when you are home, you don't need the one aimed at the porch because if the other one alerts to someone approaching you can just go to the door to see who it is, but when you are away from home, you only want to know if someone actually came onto your porch and it matters not if a child is running up and down the front walk.
2) Reduce the resolution of at least some of the cameras. Again this depends on the particular scene for each camera and what useful information can be derived from the resolution. Do you really need high resolution for all of them? Granted, we all want the highest resolution possible, but there are limits to the hardware handling of these massive amounts of data and reducing the resolution will enable more CPU time for other work.
Does each scene really provide enough information to identify a particular person? Motion detection and "Person" vs "Object" determination can be done at much lower resolution. But even at very high resolution you may not be able to tell eye color or if a face has a scar or bushy eyebrows (CSI TV shows are pure fantasy!) so you can probably get by with much lower resolution on some cameras and use the freed up CPU cycles to handle more cameras or a higher resolution on just one or two cameras where a person might be close enough for true "identification" (as opposed to Person/Object "determination").
It takes some serious thought on which scenes can really make use of high resolution and which can get by with lousy resolution and still provide the information at which SightHound excels; namely alerting when a "Person" is in a particular area of a scene, while ignoring "other" motion.
Semper Vaporo,