Hi all
We are planning to add support in our video codecs for the fast update commands defined in H.245 recommendation. This is also functionally similar to what is required by the RTP/AVPF spec (RFC4585). I would like to start the discussion in the mailing list on how to implement this so that we can do something that eventually will be defined by GStreamer and not proprietary to the codecs in our platforms. My current proposal tries to be general enough to cover what is required by both recommendations - New upstream event GST_EVENT_VIDEO_FAST_UPDATE A feedback message from decoder to encoder. Contains information of the type of feedback the decoder is producing. Event travels up to rtpbin where it is formatted and sent via the RTCP channel - New structures to be attached to the event: Name: PictureLoss Empty stuct Name: SliceLoss *Key: (type) value* First: (guint) address of the first MB(MacroBlocks) where corruption was detected Number: (guint) number of MB's corrupted PictureID:(guint) codec specific picture identifier Name: ReferencePictureSelection *Key:(type)value* Native_info: (guint*) array with a codec specific bitstring that defines the current reference picture/slices in used by the decoder Length: (guint) length of the bitstring Regards Miguel ------------------------------------------------------------------------------ This SF.net Dev2Dev email is sponsored by: Show off your parallel programming skills. Enter the Intel(R) Threading Challenge 2010. http://p.sf.net/sfu/intel-thread-sfd _______________________________________________ gstreamer-devel mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/gstreamer-devel |
Hi,
On Tue, 2010-09-07 at 14:12 +0300, Miguel Verdu wrote: > We are planning to add support in our video codecs for the fast update > commands defined in H.245 recommendation. This is also functionally > similar to what is required by the RTP/AVPF spec (RFC4585). I've already started to implement RTP/AVPF, code at: http://git.collabora.co.uk/?p=user/tester/gst-plugins-good.git;a=shortlog;h=refs/heads/avpf-timing > My current proposal tries to be general enough to cover what is required by > both recommendations > > - New upstream event > Name: PictureLoss > Empty stuct There is already a "GstForceKeyUnit" event defined for that, and it is implemented by most of the relevant open source encoder (x264enc, gst-ffmpeg and theoraenc) elements. My AVPF branch produces that GstForceKeyUnit event on the relevant AVPF RTCP messages. Documented at: http://cgit.freedesktop.org/gstreamer/gst-plugins-base/tree/docs/design/draft-keyframe-force.txt > Name: SliceLoss > Name: ReferencePictureSelection As for the two other types, I agree that we should have new events for those. Maybe extend GstForceKeyUnit for the slice loss. But ReferencePictureSelection probably requires a new custom event type. And don't feel too bad, I just re-implemented something that had existed for like a year ;) -- Olivier Crête [hidden email] ------------------------------------------------------------------------------ This SF.net Dev2Dev email is sponsored by: Show off your parallel programming skills. Enter the Intel(R) Threading Challenge 2010. http://p.sf.net/sfu/intel-thread-sfd _______________________________________________ gstreamer-devel mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/gstreamer-devel signature.asc (205 bytes) Download Attachment |
On 07.09.2010 14:33, Olivier Crête wrote:
> Hi, > > On Tue, 2010-09-07 at 14:12 +0300, Miguel Verdu wrote: > >> We are planning to add support in our video codecs for the fast update >> commands defined in H.245 recommendation. This is also functionally >> similar to what is required by the RTP/AVPF spec (RFC4585). >> > I've already started to implement RTP/AVPF, code at: > http://git.collabora.co.uk/?p=user/tester/gst-plugins-good.git;a=shortlog;h=refs/heads/avpf-timing > > > >> My current proposal tries to be general enough to cover what is required by >> both recommendations >> >> - New upstream event >> Name: PictureLoss >> Empty stuct >> > There is already a "GstForceKeyUnit" event defined for that, and it is > implemented by most of the relevant open source encoder (x264enc, > gst-ffmpeg and theoraenc) elements. My AVPF branch produces that > GstForceKeyUnit event on the relevant AVPF RTCP messages. > > Documented at: > http://cgit.freedesktop.org/gstreamer/gst-plugins-base/tree/docs/design/draft-keyframe-force.txt > > > >> Name: SliceLoss >> Name: ReferencePictureSelection >> > As for the two other types, I agree that we should have new events for > those. Maybe extend GstForceKeyUnit for the slice loss. But > ReferencePictureSelection probably requires a new custom event type. > > And don't feel too bad, I just re-implemented something that had existed > for like a year ;) > > documented so that people find it :) https://bugzilla.gnome.org/show_bug.cgi?id=607742 Stefan ------------------------------------------------------------------------------ This SF.net Dev2Dev email is sponsored by: Show off your parallel programming skills. Enter the Intel(R) Threading Challenge 2010. http://p.sf.net/sfu/intel-thread-sfd _______________________________________________ gstreamer-devel mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/gstreamer-devel |
Free forum by Nabble | Edit this page |