We've implemented new experimental motion detection algorithm that should be more accurate (less false motion detections) and consume less CPU power for processing. There are new "Motion detection settings" added in device settings section of Bluecherry admin page. 

Here you can select motion detection algorithm: 
"Default" - existing motion detection algorithm with frame downscaling feature added. 
"Experimental" - new experimental algorithm. 

"Frame downscale factor"  - higher values (up to 1.0) cause more CPU usage and higher sensitivity to small moving objects, lower values - less CPU usage and sensitivity. Each frame is scaled down by this factor before applying motion detection algorithm. Working with smaller image instead of full-size frame significantly reduces CPU power required per each camera with motion detection enabled. 

"Min. motion area %" - part of frame area space where motion happens. Motion recording is triggered only if moving object occupies more than specified percent of frame area, smaller moving objects are ignored. 

So for example if value is set to 5% and running dog occupies less than 5% of video frame, motion is not recorded, if car passing by occupies more than 5% of frame area, motion is recorded. 

"Min. motion area %" affects only experimental algorithm's behaviour, default algorithm ignores this setting. All other settings ("Frame downscale factor", sensitivity levels) apply to both algorithms.