[LAD] FW: [piksel] The future of videojack ?

[torbenh at gmx.de]

On Wed, May 07, 2008 at 01:41:30AM +0300, Nedko Arnaudov wrote:
> Juuso Alasuutari <juuso.alasuutari at gmail.com> writes:
> 
> > /*
> >   * Set the video frame rate to num1/num2
> >   */
> > int
> > jack_set_video_rate (jack_client_t  *client,
> >                       jack_nframes_t  num1,
> >                       jack_nframes_t  num2);
> 
> I think we should have this as part of control API instead. Unless we
> allow changing of sample rate on-the-fly too.

@salsaman: is it necessary to change the rate ?

> 
> > /*
> >   * Set the video processing callback function
> >   */
> > int
> > jack_set_video_callback (jack_client_t       *client,
> >                           JackProcessCallback  video_callback,
> >                           void                *arg);
> 
> I think we could reuse audio process callback.

no. video is using much more cpu time per process callback.
at a lower frequency.

30Hz vs. 44100/128 = 344.5Hz

So video needs to be done in a separate thread.
please salsaman enlighten us how you do the sync currently

i know that your patch is small and just works, but it has it
weaknesses. And while you fork, please consider this, so we can merge
at some point. Otherwise we would end with 2 different APIs.
> 
> > /*
> >   * Get the number of frames in a video port buffer
> >   */
> > jack_nframes_t
> > jack_video_get_frame_count (void *port_buffer);

i dont think there would be more than one frame per process callback.
so this call is not necessary in my opinion.

> >
> > /*
> >   * Get a frame from a video port buffer
> >   */
> > int
> > jack_video_get_frame (jack_video_frame_t *frame,
> >                        void               *port_buffer,
> >                        jack_nframes_t      event_index);

i think there should just be 2 jack_position_t in jack_client_t
and jack_video_transport_query()



-- 
torben Hohn
http://galan.sourceforge.net -- The graphical Audio language