View my account

Comments

8 comments

  • MartyG

    In theory, the distance will be the same.  In practice, in real-world conditions, it will depend upon the quality of the depth image.  The quality can be influenced by the lighting and environmental conditions of the location that the camera is in and the distance of the camera from the wall.

     

    The accuracy of the depth value is likely to be greatest in the center of the camera's view, and have the potential to reduce in accuracy at the outer edges of the camera view

     

    When using the camera with a plain wall with low texture detail or no texture, the camera's infrared emitter component should ideally be enabled so that it can project an invisible pattern of dots onto the wall's surface that the camera can then analyze for depth information.

    0
    Comment actions Permalink
  • 19829127419

    Can I understand that the depth of a point in space to the D435i camera is its vertical distance to the O-xy plane in the camera coordinate system?

    0
    Comment actions Permalink
  • MartyG

    The depth pixel value is a measurement from the parallel plane of the imagers and
    not the absolute range.

     

    0
    Comment actions Permalink
  • 19829127419

    Hello, excuse me. I would like to reconfirm with you whether the questions I asked you earlier and the answers you gave me are what I am going to express in the following content? As shown in the picture below, there is a vertical wall, and not far from the wall is placed the D435i camera, and it is facing the wall. I would like to ask whether the depth distance values of the four points P1, P2, P3 and P4 on the wall are Z1, Z2, Z3 and Z4? Also, are these four distances equal? The final question is whether the dashed lines Z1, Z2, Z3, Z4 are parallel to the Z axis of the camera coordinate system.

    0
    Comment actions Permalink
  • MartyG

    Your approach of P1 = Z1, P2 = Z2, P3 = Z3 and P4 = Z4 is fine.

     

    I used the method in the above illustration to test a D435i camera on a wall.  The Z-value results in meters for P1 to P4 were: Z1 = 0.47, Z2 = 0.44, Z3 = 0.41, P4 = 0.38.  So there was some variation on the results, which is consistent with the statement I made earlier that the depth values can drift in accuracy when moving away from the center of the camera's view - somewhere between Z2 and Z3 - towards the horizontal or (in this case) vertical outer edges of the camera's view.

     

    The values were much more consistent if the camera's Laser Power setting was increased to its maximum value of '360': Z1 = 0.367, Z2, = 0.365 , Z3 = 0.366 , Z4 = 0.365.  This is because increasing Laser Power makes the camera's invisible dot pattern projection from its IR Emitter component more visible to the camera, which benefits analysis of surfaces for depth information.  Plain walls especially benefit from a strongly visible dot pattern.

    In regard to your dashed line question, the Z-axis projects horizontally forward from the front of the camera.  So your method of parallel lines Z1-Z4 at different horizontal heights should be consistent with the Figure 2.2 official diagram above, where depth is measured from a parallel plane.

     

    0
    Comment actions Permalink
  • 19829127419

    Ok, thank you!

    0
    Comment actions Permalink
  • 19829127419

    I was wondering if you could give me a link to the official technical documentation of depth ranging for the D435i camera?

    0
    Comment actions Permalink
  • MartyG

    The official data sheet document 'D400 Series Product Family Datasheet' can be obtained as a PDF doc from the link below.

    https://dev.intelrealsense.com/docs/intel-realsense-d400-series-product-family-datasheet

    0
    Comment actions Permalink

Please sign in to leave a comment.