I am a car accident reconstruction engineer. We often use surveillance camera footage to analyze incidents. When footage is exported from a DVR system (usually in .mp4 container), sometimes it's temporally correct but sometimes the frame rates are changed, gaps in footage, frame drops, etc. In either case, I have seen the on-screen time stamp obviously differ from the frame rate in the footage, stalling, lagging, or speeding up to regain true timing. People in my field sometimes rely on the time stamp to make a speed calculation and it seems problematic to do this without knowing the issues that might affect it.
I am asking for some advice in laymans terms- what is the general process that these are written (i.e., burn-in at the camera, overlay in the DVR system, etc). And what could cause errors- truncation, slow CPU, high data rates, "saturation" of some type of buffer, and so on. I would be open to any book or training to figure this out as well. Thank you.
I have searched the internet, read some encoding books, and asked "video experts" without any luck