Does anyone know which of the camera_event_type:s best represent the actual start/stop of a video clip/stream? (if I want to be as accurate as possible, that is - I realise the difference is in the millisecond region...).
This is for videos/FITs recorded using a Garmin VIRB Ultra 30.
It seems every recorded clip starts with camera_event_type 4 (video_second_stream_start), with a timestamp that precedes e.g. 0 (video_start) - regardless of whether the "stream" is only a single clip or not.
As for the end a of clip/stream, camera_event_type 6 (video_second_stream_end) is recorded *after* e.g. 2 (video_end) in the FIT-file but consistently has an *earlier* timestamp than 2...? E.g.
[8452309] camera_event_type 2, 5033s 671ms
[8452491] camera_event_type 6, 5033s 655ms
where [xxxxxxx] is "byte" position/index in the FIT-file. ffmpeg reports slightly different numbers for the video duration than does the FIT-file.
