Hi Javier,
I have just merged your changes but now get a build failure:
[ 45%] Building CXX object
src/osgPlugins/ffmpeg/CMakeFiles/osgdb_ffmpeg.dir/FFmpegClocks.o
In file included from
/home/robert/OpenSceneGraph/src/osgPlugins/ffmpeg/FFmpegClocks.hpp:10:0,
from
/home/robert/OpenSceneGraph/src/osgPlugins/ffmpeg/FFmpegClocks.cpp:2:
/home/robert/OpenSceneGraph/src/osgPlugins/ffmpeg/FFmpegHeaders.hpp:13:38:
fatal error: libswresample/swresample.h: No such file or directory
#include <libswresample/swresample.h>
Is this additional external dependency that we now need? If so we'll need
to add these to CMake find and make scripts.
Robert.
On 4 January 2015 at 12:48, Javier Taibo <[email protected]> wrote:
> Hi Robert,
>
> Sorry for the delay. There is the new update on the ffmpeg plug-in.
>
> Now it should support any audio format that ffmpeg can manage. I tested
> it with ffmpeg versions from 1.0 to 2.4. It is using libswresample to
> convert the audio when it is in an "offending" format.
>
> New extension added to supported formats: ".ts" for MPEG-2 transport
> streams
>
> Now audio output can be requested in whatever sample rate, sample format
> and channel layout. It defaults to 48kHz, S16 and stereo when no options
> are passed to the plug-in.
>
> New parameters were added as plug-in options:
>
> out_sample_format - Output requested sample format (default S16)
> out_sample_rate - Output requested sample rate (default 48000 Hz)
> out_nb_channels - Output number of channels (default 2 channels, stereo)
> mad - Max analyze duration, time spent detecting video stream parameters
> rtsp_transport - Enable RTSP transport over udp, tcp, udp_multicast or
> http
>
>
> Cheers!
>
>
> On Wed, Dec 17, 2014 at 8:22 PM, Robert Osfield <[email protected]>
> wrote:
>
>> HI Javier,
>>
>> Sorry for the slow turnaround on this submissions. I have now merged it
>> and checked it into svn/trunk. I am still seeing problems with the audio
>> on most of the videos I have - these all have floating point planar audio.
>>
>> Since SDL1.x doesn't support floating point audio buffers I've added
>> support to the osgmovie example for compiling against SDL2.x.
>> Unfortunately this only allows us to play the audio, the actual quality is
>> very distorted. I haven't yet been able to track down why. On a movie I
>> have with S16 audio everything works nice and clearly though. I have
>> checked the addition of SDL2 support into svn/trunk.
>>
>> Could you please test out OSG svn/trunk and let me know how you get on,
>>
>> Thanks,
>> Robert.
>>
>> Robert.
>>
>> On 31 January 2014 at 17:46, Javier Taibo <[email protected]> wrote:
>>
>>> Hi Robert,
>>>
>>> I have found that since version 1.1, FFMPEG changed the way audio
>>> streams are retrieved, from packed to planar format. SDL interprets packed
>>> audio, as is used in the osgmovie example. To make the audio work when the
>>> OSGffmpeg plug-in is compiled against recent FFMPEG versions,
>>> FFmpegDecoderAudio must check for planar formats and in these cases request
>>> the samples as packed. This way all works as before. It can be checked with
>>> osgmovie example application.
>>>
>>> $ osgmovie --audio movie.avi.ffmpeg
>>> FFmpegImageStream::open audio failed, audio stream will be disabled:
>>> unknown audio format
>>>
>>> With the attached FFmpegDecoderAudio.cpp, audio sounds correctly.
>>>
>>> I am also attaching a modified version of FindFFmpeg.cmake that allows
>>> to set as FFMPEG_DIR the ffmpeg compiled in the source directory structure.
>>> It should not break anything as it only adds some additional search paths.
>>>
>>>
>>> Regards,
>>>
>>> --
>>> Javier Taibo
>>>
>>>
>>> _______________________________________________
>>> osg-submissions mailing list
>>> [email protected]
>>>
>>> http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
>>>
>>>
>
>
> --
> Javier Taibo
>
>
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
