GSoC 2011 Idea
pavelpngv at gmail.com
Thu Apr 7 23:38:45 BST 2011
В сообщении от Вторник 05 апреля 2011 20:37:09 автор Trever Fischer написал:
> > > * You actually need three parts to 'encode' a media stream: Input,
> > > encoding, and a muxer. Muxers take one or more encodings and combine
> > > them into one single output stream. An example is streaming a webcam
> > > to an icecast server. You encode the video to theora, audio to vorbis,
> > > and then combine the two streams into an ogg output via a muxer.
> > Should i create muxer as separate object? I think it's better to create
> > it as a some part of Encoder class, so user gets stream ready for
> > transmission or writing to a file.
> Yes, I feel the muxer should be separate from the encoder. Otherwise, you'd
> have to set the same container format for both the audio and video
> encoders. It makes sense to only provide one place (the muxer) for the
> developer to set which format to encode to.
> > > * Taking the above into account, the audio and video coders have some
> > > reference to a container format. The container is the responsibility of
> > > the muxer, not the encoder.
> > >
> > > * What 'properties' might be in the codecs? A defined interface for
> > > common properties isn't a bad idea, even if is a set of #define'd
> > > strings. In addition, using Qt's own property system instead of writing
> > > your own setProperties() would be a better approach. Less cruft to
> > > maintain in the long run.
> > Bitrate or sampling rate could be there. They are the properties of audio
> > codecs, I suppose. Using a Qt's property system is a good idea.
I have added Muxer class to diagram
kde-multimedia mailing list
kde-multimedia at kde.org
More information about the kde-multimedia