Since a lot of this was unfamiliar to me, I went ahead and wrote up how I worked through it in the hope that it would be useful to someone else down the road...
Here's our starting expression, let's break this down and go through it:
%{eif:trunc(t):d:2}.%{eif:(1M*t-1M*trunc(t))/10000:d:2}
eif is 'evaluate_int_format', it evaluates the expression’s value and outputs as formatted integer. From the docs:
The first argument is the expression to be evaluated, just as for the expr function.
The second argument specifies the output format. Allowed values are x, X, d and u. They are treated exactly as in the printf function.
The third parameter is optional and sets the number of positions taken by the output. It can be used to add padding with zeros from the left.... so :d:2 means a decimal integer padded to two digits.
t is the timestamp in seconds. ffmpeg appears to keep this down to millisecond accuracy, so by manipulating this we can get the output we want.
So:
%{eif:trunc(t):d:2}
trunc(t) is just the integer part of the timestamp.
the eif call here just formats that to two zero-padded digits.
(1M*t-1M*trunc(t))/10000
this is slick: 1M*t basically shifts the whole timestamp by 10e6, which gives us the total number of microseconds as an integer - or said another way we just cast the timestamp to an integer...
similarly: 1M*trunc(t) rounds down that value to the nearest second.
1.746826 seconds becomes 1,746,826 microseconds which then trunc's down to 1,000,000 microseconds.
subtracting them leaves us with just the remainder (in microseconds).
Since we want the output in hundredths of seconds, we divide by 10000 (really another shift) and this gives us the fractional leftovers, as another integer, which we then format to two digits with zero padding.
ffmpeg-utils docs
ffmpeg text expansion docs
