View my account

RealSense D450 Resolution

Comments

5 comments

  • MartyG

    Hi Andrew John Madey

    The optimal depth resolution for accuracy on the RealSense D450 is 848x480 (at 5, 15, 30, 60 or 90 frames per second), though 1280x720 is also supported (at 5, 15 or 30 frames per second).  As the D450 has fast 'global shutters' on its sensors and you wish to capture at up to 20 feet from the objects, 1280x720 and 30 frames per second is likely to be a good configuration for your particular situation.

     

    The minimum depth sensing distance of the D450 is around 0.4 meters / 1.3 feet, so the camera should be positioned at least that distance from the observed object.

     

    6 meters (19.7 feet) is the optimal distance for depth measuring accuracy on the D450.  This relates to how depth measurement error on a 400 Series camera starts at around 0 meters at the camera lenses and increases linearly over distance.  6 meters is the distance beyond which drift in the accuracy should start to become noticeable, hence 6m being the optimal depth measuring distance.

     

    At 20 feet distance, objects that are 1/2 inch / 1.27 cm diameter may have difficulty being identifiable as separate objects instead of blending into the background.  Using depth to color alignment to map a depth image onto an RGB colour image of the scene may therefore be helpful.

     

    You can use your own choice of longer USB cable instead of the 1 meter official cable supplied with the camera. It will be important that it is a high quality cable due to the high volume of data that RealSense cameras can transmit along the cable.  The company Newnex supply RealSense-validated high quality USB equipment.

    https://www.newnex.com/realsense-3d-camera-connectivity.php

    https://youtube.com/watch?v=GLQgR1jT04M

     

    In regard to alternative camera models, the requirement for up to 20 feet distance from the objects means that the D450 would be the best match for your project due to its ability to maintain depth accuracy over distances up to 20 feet / 6 meters.

     

    If you plan to purchase a D450 depth module and its required Vision Processor D4 Board separately, please ensure that you obtain the V3 version of the Vision Processor D4 as this is the correct D4 board for the D450 model.

     

    The complete-assembled D455 model which contains D450 and D4 V3 boards within its casing may also be suited to your project.

    https://store.intelrealsense.com/buy-intel-realsense-depth-camera-d455.html

    0
    Comment actions Permalink
  • Andrew John Madey

    Hi MartyG,

    Thanks! You were really helpful in clarifying the range and resolution of the camera.

    Could you go into more detail about depth to color alignment? Also, what are the capabilities of the Version 3 D4 processor?

    Thanks!

    Andrew

    0
    Comment actions Permalink
  • MartyG

    You are very welcome :)

     

    Further details about alignment can be found in the opening paragraph of the rs-align example program.

    https://github.com/IntelRealSense/librealsense/tree/master/examples/align#overview

     

    A PDF brochure about the capabilities of the Vision Processor D4 can be found here:

    https://www.stemmer-imaging.com/media/uploads/cameras/13/136603-Intel-RealSense-D4-Vision-Processor-Brochure.pdf

    0
    Comment actions Permalink
  • Andrew John Madey

    Hi Marty,

    Thanks for the help.

    So, as I understand it, the rs-align method allows me to stream either depth, or color, on top of one another in real-time.

    Would there be a way to splice the two channels after streaming, when I am analyzing the data?

    Thanks,

    Andrew

    0
    Comment actions Permalink
  • MartyG

    Alignment is the mapping of one type of data onto the coordinates of another type of data.  So rather than being an overlay of one image on top of another like a pair of separate transparent sheets, the two types of pixel are bonded together. 

     

    The RealSense SDK's rs-align-advanced C++ depth-to-color alignment example program is an excellent demonstration of this.  Because depth and color coordinates are mapped together, if the maximum observable depth range is increased or decreased then the RGB color information on the image appears and disappears as the depth information does.

     

    https://github.com/IntelRealSense/librealsense/tree/master/examples/align-advanced

    0
    Comment actions Permalink

Please sign in to leave a comment.