[MEncoder-users] Flashvideo for YouTube using mencoder

[Rich Felker]

On Fri, Feb 22, 2008 at 09:35:37AM -0500, Andy Civil wrote:
> Rich Felker wrote:
> > This is incorrect. The correct value is 30000/1001.
> 
> Rich, I'm totally NOT arguing - simply curious.
> 
> The difference between 29.97 and 30000/1001 is one in a million 
> (literally). Just to give some scale to this, if Ben used the wrong 
> value for his 16,380 frame movie, the last frame would be "mis-timed" by 
> about 1.638% of a frame. Since people frequently send out 24p content at 
> PAL rates by squishing the audio and getting away with it, this doesn't 
> seem like much.

The difference is that they retime the audio to match, and use a valid
framerate for the medium. See below..

> Similarly, if you had a TEN HOUR video, and coded the 
> video at 29.97 and the audio at 30000/1001 you would be "out of synch" 
> by a whopping (read: barely noticeable!) 36ms at the end of it.

36ms is roughly 1 frame which is generally considered to be desynced
and probably noticable with mouths and sudden sounds that should match
video (e.g. a hammer striking). Still I agree 10hr is a long time.

> So why is it such a big issue? The only possibility that I can think of, 
> is that there is a comparison in software which says "is the frame rate 
> the same" and you could falsely decide "different" when actually "same" 
> would be correct.

One issue is that 30000/1001 is a valid framerate in mpeg2 and 29.970
is not. If the movie is later transcoded, you may get errors or the
software may silently try the wrong framerate and then your DVD player
spits the disc out, etc..

It also just looks unprofessional and low quality when you get a file
and the header says some random fraction approximating 29.970 instead
of the exact value 30000/1001 (or likewise 23.976 instead of
24000/1001). It's not so much that it will cause a practical problem
for the vast majority of users, but that the wrong framerate sets off
a warning flag of "oh the person encoding this probably didn't know
what they were doing so maybe they made other mistakes like bad
inverse telecine, harmful encoder options that hurt quality, bad
filters, etc."

Rich