[FFmpeg-user] why ffplay cannot decode a sequence that can be decoded by ffmpeg

mohammed bey ahmed khernache mohbeyinfo at gmail.com
Tue Jun 12 10:45:33 EEST 2018


>  In this case, ffplay is the wrong tool for you, ffmpeg -i input -an -f
null -
> may be a better idea.
FFmpeg decodes a video depending on the hardware capacities, i.e., a video
of 10 sec may be decoded in more than 1 mn or in only 5 sec. Therefore,
FFmpeg doesn't incur miss rate (all frames are decoded and no one is
dropped).
To be clear, in my experiments, I want to evaluate video decoding using
FFmpeg while running on multi-core.
*Question:*
Is there a way to decode a video using FFmpeg at a certain fps (e,g,, 25
fps) independently of the hardware capacities (*m**aybe by adding some code
to FFmpeg and rebuilding it*). Of course, we will get different qualities,
i.e., for example using 2 cores we will get 100 dropped frames whereas
using 4 cores we will get only 5 dropped frames.

>  Frames per second is
> available from ffprobe, right?
Yes, it gives the fps at which it is encoded. In my experiments, I need the
fps while decoding the video. It is different because it depends on the
hardware performance.
>  -f null /dev/null
as I understood, this parameter allows to throw the decoded frames, i.e.,
the decoded frames are lost.

​Thank you again for all your replies.​


On Tue, Jun 12, 2018 at 3:27 AM Zak <ffmpeg-user-email at m.allo.ws> wrote:

> On 2018-06-11 at 5:32 PM EDT, Carl Eugen Hoyos wrote:
> > 2018-06-07 23:40 GMT+02:00, mohammed bey ahmed khernache <
> mohbeyinfo at gmail.com>:
> >
> >> My purpose is to calculate some metrics such as: fps, miss rate, etc, of
> >> video decoding. So I need only to decode a video without displaying it.
> >
> > Sorry for missing this:
> > In this case, ffplay is the wrong tool for you, ffmpeg -i input -an -f
> null -
> > may be a better idea.
> >
>
> Okay, I am going to ask the questions that have been on my mind for this
> entire thread:
>
> 1. What is the goal of the original email? Frames per second is
> available from ffprobe, right? Does "miss rate" refer to cache misses
> while operating on the compressed video data? Won't the number of cache
> misses depend on many factors, possibly including whether the computer
> hardware is also busy rendering and displaying a video on the screen,
> such that disabling all video display actually will change the "miss
> rate"? This does not seem like a characteristic of the video file
> itself, it seems like a statistic that is only available during a
> specific instance of playing the video with specific software and
> hardware. Or maybe I am misunderstanding. The number of cache misses may
> in practice be exactly the same every time you play a given video on
> certain hardware, but the original email says the test is being run
> remotely on a computer that cannot display video, so any actual playback
> will occur on a different computer that will have different performance,
> unless I am misunderstanding.
>
> 2. What does ffmpeg with -f null really do, and how is it different than
> ffprobe?
>
> Thank you,
>
> Zak
> _______________________________________________
> ffmpeg-user mailing list
> ffmpeg-user at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-user
>
> To unsubscribe, visit link above, or email
> ffmpeg-user-request at ffmpeg.org with subject "unsubscribe".


More information about the ffmpeg-user mailing list