Greetings. Hope it's not too late to glean some insight and perhaps help shed some light on this MP4 issue...
I’m working with an array of various models of Reconyx camera traps in a Kenyan forest. The XR6 Ultrafire produces both JPGs and MP4s that can be captured for each episode, i.e., one JPG followed immediately by a video. In Finder, the images sort correctly on the SSD fresh from the camera. After ingest by PM, the MP4s all have a “Capture Time” that is 3 hrs greater than the true capture time; the JPGs have the correct local time, which in Nairobi is UTC+3. Thus, when sorting the batch on the default “Capture Time” in the contact sheet, the JPGs and MP4s are all badly out of sequence making the export of precise time-series summaries impossible. (NB: *.avi files from another camera model, Hyperfire 2 Pro HP2XODG, don't exhibit the anomaly.)
The timestamp anomaly occurs apparently because PhotoMechanic changes the "capture date" on ingest. I did an ingest test with my iMac system date set to UTC-0, that is three hours earlier than the actual Nairobi time, UTC+3. The batch ingested were in the correct time sequence when sorted on PM’s default Capture Time option, unlike a batch ingested with the iMac internal clock set back to Nariboi time, UTC+3: all the latter MP4 Capture Times were increased by 3 hours.
It gets more complicated. I tried a custom sort of the date-corrupted UTC+3 batch using the {datesort} variable. The sort rendered both JPGs and MP4s in the correct time sequence, despite the fact that the IPTC data of the MP4s continuing to show the +3 hour anomaly in the IPTC data.
So question: what meta-data is PM sorting on when using {datesort}? It’s important to know which variable has the actual time because we subsequently use those times, measured in seconds, exported into Excel tables in order to determine which bursts of photos constitute one capture event.
Many thanks if you managed to read this far.