carcachute

Have u try Gigabyte Brix ? I think its better then Intel NUC I found out that the very latest version of ZoneMinder can use the low-res feeds for analysis. Sounds like the pulling-from-cameras architecture (as opposed to pushing to the server) is the way to go. I think I'll give it a shot and upgrade to a fancier box if it is too much of a burden. The Intel NUC looks amazing, just the kind of thing that could easily be locked into a vented cabinet someday.

Hello! I have a simple setup with four cameras: Three D-Link DCS-942L cameras (640x480 @ 30fps) monitoring the garage and main living area One D-Link DCS-2230 camera (1920x1080 @ 15fps) looking outside (through a window unfortunately). All four record continuously to 8GB SD cards and feature (primitive) video motion detection and PIR sensors. I use the PIR on the inside-facing cameras and video motion on the outside-facing camera. The initial security need that caused me to buy them has passed but I've come to enjoy the peace of mind when I'm away, so I have kept the cameras. The setup generally works okay except: The SD cards hold only about 1.5 days worth of recording (the old files get purged automatically) The physical security of the recordings is very poor--all someone has to do is take a camera The outdoor-facing camera has poor daytime image quality and nonexistent night vision because it looks through a window, and it would be nice to have the night vision. The outdoor-facing camera has very simplistic motion detection that false triggers frequently when the sun goes behind a cloud or the trees blow in the wind. On some days I can get 200 e-mails. I'd like to replace one of the cheaper cams with something nicer like the HikVision 2032 3MP bullet and actually mount it outside, but the HikVision (and most other) cams require some sort of server. Gathering dust in the closet is an Intel Atom D510 (dual 1.6 GHz) mini-ITX board and case with 2GB of RAM. All I need to do is add a hard disk and small SSD and replace a fan, and I have a nice low-power machine for Linux or BSD. I am considering whether to rehabilitate that system and set up a server for recording, but I want to know if my approach is feasible. I am aware of three general approaches to capturing video with IP cameras: What I currently do, using on-board storage on the cameras. Unfortunately most of the nicer cams like HikVision 2032 don't store anything internally. Using a NAS (FTP, NFS, SMB) and having the cameras push video files to the server. My cameras will send only short (10-15 second) clips that may not actually capture the subject, because the PIR has a wider field of view than the camera lens. There is a delay before the camera can record again, which causes gaps between motion clips. It seems IP cameras in general can upload video clips in response to motion detection (which itself is a bit spotty and prone to false positives) but cannot push uninterrupted video to a NAS. Using software such as BlueIris or ZoneMinder to pull video feeds from the cameras, analyze the video for motion. This would provide central management and (I assume) longer pre- and post-event record times, or continuous recording if I want. The problem with this is that it requires massive processing overhead and ignores a lot of the processing available in the cameras already. I'm leaning toward approach #3, pulling video from the cameras to a database-driven system. However, I'm concerned about scalability. I've found posts on this forum and others where folks have had just a few cameras run a quad-core i7 into the ground. On the other hand, there are posts like this one (http://www.cctvforum.com/viewtopic.php?p=250628#p250628) where someone is successfully using an Atom box. I'm trying to account for the difference. The D-Link cams have onboard PIR sensors. Can software such as BlueIris or ZoneMinder use that signal as a trigger, saving the overhead of video analysis? They also can produce low-resolution (for example 320x200 @ 10fps JPEG) streams and full-resolution H.264 streams simultaneously. Can currently available software take advantage of this to reduce overhead? What I've read is that ZoneMinder uses Motion, which decodes the stream, analyzes it, then re-encodes it. I found on the ZM forums that you can set it to skip frames for motion detection, which helps, but it seems like using a low-resolution feed for motion would be much more efficient and allow the system to scale better on low-power hardware. Am I right in thinking that it should be possible to do what I want with an Atom box? Are packages like ZoneMinder configurable enough to make my IP system scalable? Thank you!