Closed Me817 closed 2 years ago
Hi @Me817 the frame rate of the received data also depends on the CPU and system load. With the same settings on the same system, what do you get directly with the Kinect SDK (k4arecorder.exe
)?
Closing due to lack of response. Please reopen if you would like to investigate this further.
Hello,
Describe the bug I recorded a mkv-file with the azure_kinect_recorder.py with the
default_config.json
: that means a sampling rate of 30 fps. The length of the video is 57 seconds ("stream_length_usec" : 57000011
). If I understand it right I would have 57x30=1710 frames in this mkv-file. But I only get 394 color and depth images and also 394 point clouds. How cloud that be?To Reproduce Steps to reproduce the behavior:
azure_kinect_recorder.py
without giving a specific config-file using thedefault_config.json
.python path/to/azure_kinect_mkv_reader.py --input config_nfov_record.mkv --output frames
Expected behavior I would expect that I get 1710 jpg and 1710 png files and out of them 1710 pcd files if I have a stream length of 57 seconds and a sampling rate of 30 fps. Does
azure_kinect_recorder.py
sample with 30 fps if it is configured in thedefault_config.json
? How can I achieve a sampling rate of 30 fps?Screenshots If applicable, add screenshots to help explain your problem.
Environment (please complete the following information):
Additional context Why are the timestamps of the color and depth respektively the point cloud not equidistant? Bewteen the first and the second point cloud is a difference of 66666 usec, between the second and the third 133323 usec, and between the third and the fourth 266689 usec. Between the last and second last 133334 usec and between the second last and the point cloud before 100011 usec. But why?
I would be very thankful for a hint what I can do. Thank you in advance.