[MEncoder-users] Want a sample command to convert files to FLV

The Wanderer inverseparadox at comcast.net
Fri Aug 18 13:18:49 CEST 2006


howard chen wrote:

> On 8/18/06, The Wanderer <inverseparadox at comcast.net> wrote:
> 
>> howard chen wrote:

>>> yes, after i use "-v", i can get the average fps now...
>> 
>> What's in your ~/.mplayer/config file? If the 'total time' and
>> 'total frames' are not printed by default, then there's something
>> weird going on.
> 
> seems nothing (it is the default?)
> 
> # Write your default config options here!

That *is* strange, then. Not sure where to go from there.

>>> in fact, i wonder the need to playback the whole file to get the
>>> average fps, even the player play the first 20-30 seconds could
>>> get the average fps.
>> 
>> That will give you the average FPS over that small chunk of the
>> file, but it won't necessarily be the same as the average FPS of
>> the entire file; if the frame rate changes later on, the average
>> FPS will be different.

(This seems so self-evident to me that it seems hard to understand how
you are managing to disagree with it...)

>> In point of fact, I noticed that exact effect in the course of
>> testing to make sure that the numbers I expected were in fact
>> gettng printed. Once I let the video finish 'playing' and got total
>> time of 420.2 seconds with 11822 frames, for an average FPS of
>> ~28.13422; another time, I interrupted it early on at 35.0 seconds
>> (by coincidence) and 832 frames, for an average FPS of 23.77143.
>> Considering the desync I sometimes get in non-variable-FPS files
>> whose FPS value is off by (as far as I can tell) about 0.1 frame
>> per second, I'd say that 4.36 frames per second is a very
>> significant difference.
> 
> what kind of file you are using? seems as RC said, only WMV use
> variable video bit rate, i doubt the difference in FPS is so large?

A WMV file - officially reported as being 1000.000 FPS, which is the
value MPlayer gives when the frame rate is variable.

(I'm assuming you meant "variable frame rate", because WMV is certainly
not the only format which uses VBR.)

> the ffmpeg can determine the average fps of my wmv file during
> encoding, so i doubt the need to play the whole file in order to get
> the average fps.

"Average FPS" is, by definition, "number of frames per unit time". If
there is a way to know both of those numbers for the entire file without
analyzing or otherwise processing the whole thing, then the average FPS
can be determined without having to do that; however, I know of no way
for that to be possible for WMV. I don't know where FFmpeg gets its
reported input frame rate from, but for the source file I was using it
says

==
   Stream #0.1: Video: wmv3, yuv420p, 320x240, 30.00 fps(r)
==

and, if I actually let it transcode the file and then play the result in
MPlayer, it does not have the same total time as the original file -
which, to the best of my knowledge, means that the frame rate is
different.

If I just remux the file with FFmpeg, it reports that there are 11823
frames constituting 418.8 seconds (average FPS: 28.231). If I transcode
the video with default settings (i.e. msmpeg4, minimal bitrate, etc.),
it reports that there are 12565 frames constituting 418.8 seconds
(average FPS: 30.000).

MPlayer thinks that the remuxed file has 11822 frames constituting 418.2
seconds (average FPS: 28.269), and the the transcoded file has 12564
frames constituting 420.8 seconds (average FPS: 29.857) as well as a
flood of MV errors which probably deserve a bug report somewhere.

As you can see, there are an excess of different numbers; the only ones
that agree with each other are FFmpeg's 'estimation' of the frame rate
and the output-frame-rate value it uses when told to transcode the file
(which, since it's based on the other, doesn't mean much). Given that, I
think I'm inclined to trust the math here.

-- 
       The Wanderer

Warning: Simply because I argue an issue does not mean I agree with any
side of it.

Secrecy is the beginning of tyranny.



More information about the MEncoder-users mailing list