The matter of fact is that compressed video (easy explenation) “drops” frames along the way, which need to be re-generated by the computer that plays back the video. Best case is that the compression drops one frame out of tree. A-B-C // B is dropped. Which means that upon playback the droppped frame (B) is re-calculated by the computer using the information of A and C. The more a video gets compressed, the more frames are dropped and the more computing power is needed to restore (de-compress) the video. This can result in delays.
What’s more, when nudging video from frame to frame, you inevitably will freeze on a “re-calculated” frame, which is never accurate.
So placing -for example- a doorslam can -and will- result in an inaccrate placement of the impact.
I’ve seen cases where, after receiving a HiRes video, we were 3 frames off versus the mp4 version they send earlier.
The reason why DnDx, ProRes and Photo Jpeg are recommended, is that these are the only codecs which preserve the frame-by frame continuity.
No frames are dropped. Hence why they are “big”.
Everybody working in Audio Post, please make it a habit to send delivery specs to the video people.
Specs that include BITC and 2pop.
It will save your ass on almost every project.