Thanks therealdavidp, I'm a bit delayed, but will test your options. My guess is none work - let's see (and hope) I'm wrong.
A quick response to your more general point - indeed time stamps from metadata are nearly useless because they are often not reported (e.g. 0, 8.333ms, 16.666ms, 50.000ms...) so it's not obvious how to align when skips occur (e.g. 50ms time stamp) to which video frame in the .h264 or mp4 stack. However, I have found that the camera encoder (?) robustly ticking at 120FPS - even if the meta-data is missing. But even so there is variance in the # of encoded frames reported (e.g. 7204 for a 60sec video is the most common value). It might actually be easier to align the frames by recording system time at start/end of recording and aligning to this, but often it wont' give 8.333ms frames.
I'll report more shortly.
A quick response to your more general point - indeed time stamps from metadata are nearly useless because they are often not reported (e.g. 0, 8.333ms, 16.666ms, 50.000ms...) so it's not obvious how to align when skips occur (e.g. 50ms time stamp) to which video frame in the .h264 or mp4 stack. However, I have found that the camera encoder (?) robustly ticking at 120FPS - even if the meta-data is missing. But even so there is variance in the # of encoded frames reported (e.g. 7204 for a 60sec video is the most common value). It might actually be easier to align the frames by recording system time at start/end of recording and aligning to this, but often it wont' give 8.333ms frames.
I'll report more shortly.
Statistics: Posted by catubc — Thu May 08, 2025 5:05 pm