GNOME Bugzilla – Bug 651806
Implement Video compositing and Audio Mixing
Last modified: 2013-09-24 16:03:38 UTC
Until now, there is no way to mix audios/videos from various layers. We need to get something with a similar behavior as what now have in pitivi.
The solution in pitivi is:
* Have mixer elements smartVideomixerBin and SmartAdderBin (in pitivi/elements/mixer.py)
* All tracks add this mixer in a gnloperation with a priority of 0 and which is expandable so it is in the whole duration of the track so any trackobject added to track is plugged on the mixer in pitivi/timeline/track.py
This result in:
+ layers (audio/video from different layer) are mixed when the alpha channel of objects on higher layers are < 100
We also currently have an optimization that take advantage of the alpha passthrough capabilities of the videomixer, we should try to keep it.
If we want to reproduce something similar, we re-implement those custom GstElement-s and modify ges/ges-track.c to handle it.
In GES, we should use the new video mixer: videomixer2
Also I guess this is something we might want to handle directly in GnlComposition, passing a videomixer or adder to it?
Looking into it/Working on it :)
Currently, this is one of the key regressions in Pitivi. As nobody is working on it at the moment, unassigning.
This bug has been fix between
Author: Mathieu Duponchelle <email@example.com>
Date: Mon Jun 3 23:02:15 2013 +0200
ges: Add a framepositionner element used in ges-smart-mixer and ges-uri-source
It adds metadata on the buffers and the mixer parses them.
This is done because we want to keep positionning properties
and set them on the dynamic mixer pad.
Author: Thibault Saunier <firstname.lastname@example.org>
Date: Sat Mar 30 19:02:52 2013 +0100
track: Implement infrastructure for mixing