[Video] Recording premix?

[Carl Karsten]

On Fri, Jun 21, 2013 at 10:04 AM, Paul Wayper <paulway at mabula.net> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi all,
>
> One of the downsides of DVswitch has been that it only allows you to record
> the final output, after the operator has chosen which of the many video (and
> possibly audio) feeds to display.  On the one hand, this can lead to 'Mark
> Pesce' moments where it would be really great afterward to display an
> audience reaction or a picture of the presenter rather than the (offensive)
> slide.  On the other hand, there are many times where the operator just
> doesn't quite start the recording on time, or switches to the slides a bit
> too late, or cuts away from something a bit early.
>
> (Aside: one of the problems I had with LCA 2013's video was that one person
> in the organising team decided that, in order to prevent any 'Mark Pesce'
> moments, all rooms had to record raw streams for the camera and the
> presenter's laptop as well as the final output.  This caused a lot of
> problems, both in engineering a solution and in trying to store all the
> excess video.)
>
> Even with raw DV, hard disk sizes these days make it not infeasible to
> record all streams.  Certainly once we look at storing compressed video, the
> space requirements to store the raw video streams cease to be a problem.  We
> simply need to store a relatively simple cut list file - such as what Melt
> uses - as well as the video.  Transitions such as fades, and picture in
> picture display, can also be stored relatively easily (although I have no
> idea how to do such things with Melt).
>
> Reviewing and editing those cut lists would be a matter of time and
> opportunity.  Obviously, if no further work needed to be done the project
> could simply be mixed down and encoded as is.  But the opportunity to fix
> things is always useful.  It would also make sense to have the ability to
> only record the post-mix anyway - to render on the spot, as currently done -
> if the user was space or time constrained or had other reasons.
>
> Thoughts?

Nice to have, but we have lived this long without it so it obviously
isn't a show stopper.

Right now the videos I make of the R-Pi talks are crap because I can't
capture the hdmi.
Also not a show stopper, but way higher priority, so I think we should
focus on getting gst-swtich functional, and then add this feature to
it.

-- 
Carl K