View my account

Frame loss or multiple

Comments

3 comments

  • MartyG

    Hello, it is the nature of bag frame extraction that some of the frames will be missed.  You may be able to extract most of the frames, but usually not 100% of them.  The number of frames extracted may change each time the extraction process is run.

     

    If you are extracting using a program script then setting a bag playback instruction called set_real_time to false can improve stability if you are not using this instruction already, as described at the link below.

    https://github.com/IntelRealSense/librealsense/issues/10312

     

    Repeated frames could occur if frame drops are taking place or if a particular frame is decided by the RealSense SDK to be bad or incomplete, causing the SDK to return to the last known good frame and move forwards again from that point.

    0
    Comment actions Permalink
  • 3526810272

    If the first extraction is normal and an image with a depth of 16 is extracted. Is it difficult to see frame drops at depth_10 for the second time? Will the frame at depth_16 be the same as the one extracted for the first time? In other words, will the second frame_number==10 be lost or replaced by the image with frame_number==11

    0
    Comment actions Permalink
  • MartyG

    The depth values recorded in the bag will remain constant.  An extracted frame with a particular frame number should also be the same frame of that number that is stored in the bag.

     

    However, it is not guaranteed that a specific frame that was extracted on the first run of the extraction program will be extracted on subsequent runs.  It is not possible to predict in advance which frames might be dropped.

     

    The risk of frame drops occurring can be reduced by increasing the frame queue size value in your code.

    https://github.com/IntelRealSense/librealsense/wiki/Frame-Buffering-Management-in-RealSense-SDK-2.0#latency-vs-performance

    0
    Comment actions Permalink

Please sign in to leave a comment.