Multiple Depth Sensors vs Web Cameras vs Action Cameras Comparison

From iPi Docs
Jump to: navigation, search

There is no straightforward answer to the questions which is better — Multiple Depth Sensors or Multiple Web Cameras or Multiple Action Cameras configuration. The accuracy of tracking is almost the same. Pros and cons summary:

Cameras/Sensors Pros Cons
Multiple Depth Sensors (2 to 4 depth sensors)
  • Easier to set-up and calibrate
  • Require less space due to wider angle of view, minimum required space is 7 x 4 feet ( = 2.2 x 1.2 meters)
  • No requirements to clothes color
  • Capture area is limited with 7 x 7 feet ( = 2 x 2 meters)
  • Framerate is only 30 fps (more tracking errors on fast motions)
  • More problems with occlusion (especially for complex motions including rotations)
  • Kinect 2 (Kinect for XBOX One) sensors require a separate PC for each sensor for recording
  • Cannot be used outdoors
Web Cameras (3 to 16 cameras)
Sony PS3 Eye or Logitech C922
  • More accurate tracking for head and feet (in comparison with depth sensors)
  • Bigger capture area (in comparison with depth sensors), up to 20 by 20 feet ( = 7 by 7 meters)
  • 60 fps framerate allow to track fast motions
  • Less problems with occlusion (especially for more than 6 cameras)
  • Allows outdoor capture (not in sunny weather though)
  • Relatively low price
  • Set-up and calibration takes more time
  • More issues with USB controllers (you need 3 USB controllers for 6 cameras, and 4 USB controllers for 8 cameras)
  • You need more space, minimum required space is 13 x 13 feet ( = 4 x 4 meters)
  • Higher requirements for HDD speed
  • A lot of USB cables on the floor
Action Cameras (3 to 16 cameras)
  • No need in PC and external power supply for recording (high mobility)
  • No USB cables on the floor
  • Wider view angle (FOV) results in bigger capture volume (in comparison with web cameras)
  • Wider view angle allows to record in smaller rooms (in comparison with web cameras)
  • 90 and even 120 FPS (depends on the model of action camera and lighting conditions)
  • Intrinsic parameters of each camera must be determined in advance (camera optics calibration)
  • A few manual and time consuming steps are required to get multi-camera synchronized video:
  • Right now there is no simple way to use Motion Controllers with action cameras

If you choose to go with depth sensors, see also the comparison of different models.

General Recommendations

  • Dual Depth Sensor configuration is very practical, it is easier to use and able to track many kinds of motions. We can recommend it if:
    • Your room is not very big (less than 20 x 20 feet = 6 x 6 meters)
    • You do not need to track fast complex motions like energetic dancing, acrobatics, etc.
    • You do not need to track 2 actors
  • Multiple Web Cameras (Sony PS3 Eye or Logitech C922) is more powerful option, but you have to invest more time in learning. We can recommend it if:
    • 7 x 7 feet ( = 2 x 2 meters) capture area is not enough for your motions, and your room is bigger than 20 x 20 feet ( = 6 x 6 meters)
    • You plan to track fast complex motions like energetic dancing, acrobatics, etc.
    • You plan to track 2 actors
  • Multiple Action Cameras configuration is almost similar to multiple web cameras configuration but involves more tricky workflow. We recommend it if:
    • You already have required set of action cameras
    • You want to be extremely mobile
    • You're going to capture very fast motions at 90+ FPS