Motion Detection

You can trigger the shutter, start a script or record video by detecting motion in live view.

Note: The motion trigger works only when using the non-GPU live view mode.

Note: The following video tutorial shows the pre-5.4 method for setting up motion detection.  The controls in the Motion tab have changed in 5.4 a lot.  You can still use the old tutorial video below to get an idea how it works, then read further below to see how to set up the new motion detection controls.

The goal is still the same – get dots in a target to capture an image.

And here’s the new Motion tab from 5.4.

Setting the Trigger

  1. Turn on live view and open the Motion tab.
  2. Reset the slider values back to defaults.  You can do this by clicking on the small button on the right side of the slider.
  3. Set Dots/Enabled to checked.
  4. Set Filter/Enabled to checked.  This will display a small image in the upper-left-hand corner of the live view display.  This small image is the image that will be used for detecting motion. It is basically the live view image with a grayscale, Gaussian blur and a Sobel contour detection filter applied.
  5. Move the Filter/Brightness slider to the left to darken the small image.  You want a fairly dark image, but you need to be able to clearly see the edges.  If it is too bright you will get too much noise.  Moving it to a position of 60 usually works for most scenes, but lighting variations will require you to change this setting.
  6. Move the Rejection slider to the smallest value where you do not get any dots appearing due to noise.
  7. Now try moving something in the scene to see how it handles motion.  If you do not get enough motion dots due to real motion, reduce the Rejection slider value further.  If you get dots caused by noise, increase the dot size until the noise-caused dots go away.  Now try the motion again and it should be ok. The Dots Size cause noise to be averaged over a wider area so it helps for reducing noise.  Small dots (less than 20) tend to handle noise poorly.  Version 5.3 dots were only size 13.  Size 30 is a good place to start, though I usually wind up with size 40 or 50. The Buffer Age can be increased if noise is still a problem.  This helps to average the image stream over time, however the higher the value, the more motion ghosting you will get.  1000 seems to work good, though less than 500 causes problems on slower computers when flipping to different tabs or in and out of live view (it will detect it as movement due to a slight lag in live view image transmission).
  8. Once it looks good, you can disable the filter if you want and you are good to go.
  9. Draw a target around the intended area in which motion dots will appear. To draw the target, put a checkmark beside the ‘Draw Target’ checkbox and then then left-click on the image to draw corners of the target. At least three corners are needed to define a target. You can clear the target by right-clicking on the image, or by clicking on the ‘Clear’ button.
  10. Reduce the ‘Trigger on X Dots in Target’ slider until the target starts to turn red. The target turns red when at least X – – Dots are visible in the target area. If the Action is enabled, triggering would be performed.
  11. Set the Action reset to 10 seconds. Set it high enough so that the trigger does not continually fire, unless you want it to do so.  Leave enough time for the captured image or movie file to be transferred.
  12. Enable the Action. The trigger is now live. If the number of dots meet or exceed the X Dots settings, the trigger will fire the action specified in the action list.

Tips

  • The motion detection image analysis requires a fairly fast computer. Intel processors are able to process this faster than cheaper AMD processors (except Atom CPU’s, which are still fairly slow). Slower computers will display a choppy/laggy live view image stream, making image detection more difficult. Try using the turbo mode.
  • There is a delay of several hundred milliseconds from when when the motion is detected to when the shutter is triggered. This means that fast moving objects may be difficult to capture. Water drops, for example can’t be captured with this method. However, you can experiment by moving the target area to an position in advance of the actual area where you want the subject to be when captured (lead it a bit).
  • Accurate motion detection is dependent on good lighting and contrast between the subject and background.
  • The motion detection also works while live view zoom is enabled.
  • You can use burst captures, however you will need to disable the body autofocus as in a burst, the first shot is captured with the current focus, then the subsequent shots will automatically try to phase focus for each shot.  To get around this firmware limitation, disable the autofocus on your lens or body. Or, set the phase focus box in the view find to the position where the subject will be upon image capture.