What’s the best way to save the Azure Kinect’s 4K color video, while simultaneously saving the Kinect CHOP joint/body positions into a .chan file? I’m working on a data science project and the goal is to record at the highest FPS for the Kinect with no dropped frames in the video and no duplicate joint data in the .chan. We’re going for recorded accuracy and are looking to have a row of joint data in the .chan associated with each frame of the video.
I currently have the ‘Sync Image to Body Tracking’ turned ON, but I’m running into the problem that the Kinect joint data seems to hang up and write duplicate / many repeated rows of data into the .chan file. Also in the movie file TOP out, I have the fps set to ‘me.time.rate’ to match the current fps of TD. (note when Realtime is on I’m getting ~58-61 fps)
One thing to note is that the Azure Kinect hardware maxes out at 30fps, but I get the smoothest recorded video either running with real-time off, or at 60fps with real-time on.
My intuition would be to keep real-time off, but we get tons of repeated data in the .chan file.
Any pointers would be greatly appreciated!