[FFmpeg-user] Screen capture without sound
thilo.borgmann at mail.de
Fri Mar 4 19:12:54 CET 2016
Am 04.03.16 um 17:51 schrieb 桃源老師:
> Hello Thilo-san, Thanks for reply.
>> 2016/03/05 0:58 A.M. Thilo Borgmann <thilo.borgmann at mail.de> wrote:
>> Am 04.03.16 um 12:28 schrieb 桃源老師:
>>>> Please test without the scale filter (only crop) or
>>>> reduce the screen resolution.
>>> Unfortunately, with above condition, I can't get good audio.
>>> I'm now thinking desktop recording with ffmpeg is not possible yet...
>> The avfoundation device is under reconstruction. It includes cooperation
>> with several people and it seems it is not on the edge of being finished.
> OK. It's interesting information.
>> For the time being you might checkout "avf2" or "avfarc" branch on my
>> github repo:
> I'd like to confirm my git operation is correct or not. Here's console log. Would you please check them?
> $ git clone https://github.com/thiloborgmann/FFmpeg.git
> Cloning into 'FFmpeg'...
> remote: Counting objects: 438270, done.
> remote: Compressing objects: 100% (8/8), done.
> remote: Total 438270 (delta 3), reused 0 (delta 0), pack-reused 438262
> Receiving objects: 100% (438270/438270), 171.74 MiB | 2.03 MiB/s, done.
> Resolving deltas: 100% (342404/342404), done.
> Checking connectivity... done.
> $ cd FFmpeg/
> $ git branch -a
> * master
> remotes/origin/HEAD -> origin/master
> $ git checkout -b avf2 origin/avf2
> Branch avf2 set up to track remote branch avf2 from origin.
> Switched to a new branch 'avf2'
>> The avf2 implementation is far superior to what is currently in FFmpeg
>> but I cannot guarantee that all your problems are solved.
> I will try.
What you've done so far seems to be correct to me.
You can test if you've done it right by
$ git checkout avf2
$ git log -2
In the output there should be the commit message "major overhaul".
You can compare the output with the commit history for the avf2 branch
in the web interface of my repository.
If you see the same history on the web as in your local repo clone,
everything should be fine.
>> If needed you should easily be able to rebase all the patches there to
>> current FFmpeg git HEAD.
> I clone your mirror. (Am I understand correctly?)
No. This is a bit more complex and you should look it up in the scm git
book ("working with remotes" and "rebasing") instead of getting
potentially wrong git commands from me.
>> p.s. Audio frame drops happen more often the more video processing you
>> do during capturing. Thus all video processing should preferably be done
>> after capturing.
> OK. Then should I capture raw video and raw audio?
> What container should be used?
Either use rawvideo if the resolution is not too high or use a very fast
and "simple" video encoder - like mpeg2, maybe mpeg4. Audio can be kept
in raw format for the time being.
I use almost the same hardware btw and it is possible to get "good"
recordings out of it. They will even be better after reconstruction is
I appreciate any feedback on this. Please also consider to open a ticket
on trac (there are some about audio stuttering already which should be
your problem), if you have reproducible issues. If there are issues with
avf2 then do not use trac, please mail me directly in that case.
More information about the ffmpeg-user