• TI thinks resolved

OPT8241-CDK-EVM: the opt8241 depth data distribution like circle?

Part Number: OPT8241-CDK-EVM

Dear Larry&Anand,

        I have used  a different color represent depth data. just like the below figure.

        It's very interesting the depth distribution like circle. this phenomenon affected the RGB-D rectification

        Can you give me advise about this. 


  • Hi Larry & Anand,

    Can your give me advice about this question?


  • In reply to None_feiyu:

    From what I see, it can be explained by turning off calibration and having a white board in front of the camera at a big distance. I am assuming you are using a color coding for the depth, is that true?
  • Hi None,

    If you are plotting the raw phase data or the depth data from the SDK, this represents the radial distance. Therefore the image you get for a properly calibrated camera will look like concentric circles - this is expected behavior.

    If the lens parameters are known, the SDK can also provide point cloud data. The Z values from the point cloud should give you a flat surface when plotted.
  • In reply to None_feiyu:

    Your color values are repeating, that tells me your value-to-color mapping is aliasing, most likely caused by not matching the full range of values to full range of colors. Please double check.
  • In reply to Anand Gadiyar:

    Dear Anand,

    I don't understand why this is expected behavior? I think it should be Linear gradient color?

    hopes you give me more advise.


  • In reply to Larry Li:

    Dear Larry,

              Thanks for you information.

              What I'm concerned about is the color map like Concentric circles,just like below figure.


  • In reply to None_feiyu:

    Anand's answer is on mark. The depth map measures the distance from the pixel the point in the scene it images. In the case of a flat wall, the center pixel distance will be the shortest, and increasing for pixels further away from the center. The reason for the repeating color is likely a bug in your depth-to-color mapping--which can happens when input range is not correctly mapped to the color range.
  • In reply to Larry Li:

    Also please make sure your image is not saturating (too strong of reflection).
  • In reply to Larry Li:


    Is this issue resolved? I think someone from your end was supposed send us some raw data VXL file with the above issue to review.

  • In reply to Larry Li:

    Dear Larry,

    Sorry for the delay reply. the email sent for you is from me.

    We have understand why the depth data look like concentric circles,but this issue affected the RGB-D rectification

    Now this issue is suspend,we have no idea about this issue. because in the RGBD Piond cloud it look like ok.

    if want to solve this issue fully,it mybe need two RGB sensor for the RGBD rectification.

    thanks & regards.