Then the video was replayed and additional synchronization points were added to the list for later update.
Surprisingly we had different experience than in the first study with drift of video and log file records.
There was no drift of times through the whole video, the difference at the end of the 10 minutes video was below 2 seconds.
We assume that the synchronized visualization of log and video allowed more precise localization of events in both sources, which removed mismatch generated when
switching between video player and excel spreadsheet with log in first study.
Therefore average time for log video synchronization was approximately 2 times faster,
see Table 1. When the audio/video recording and log file were synchronized, annotation editor view with direct video rewinding controls saved us time when we were repeatedly inspecting important or interesting parts of video.
It took less then 2 times the length of the video to annotate the video with observer annotations and with new annotations based on video replaying, comparing to approximately 3 times the video length in the first study, see Table 1.