[Bug 627459] theoraenc should provide option for TH_ENCCTL_SET_DUP_COUNT

GStreamer (bugzilla.gnome.org) bugzilla at gnome.org
Mon May 9 03:13:31 PDT 2011


https://bugzilla.gnome.org/show_bug.cgi?id=627459
  GStreamer | gst-plugins-base | 0.10.x

--- Comment #21 from Sebastian Dröge <slomo at circular-chaos.org> 2011-05-09 10:13:27 UTC ---
(In reply to comment #20)
> Ok, i almost give up :)
> 
> > (In reply to comment #19)
> > > IMHO, The latency is imaginer, because ogg do not use timestamps for video
> > > stream. We need only correct count of frames per second.
> > 
> > For creating files the one-frame latency is fine. Sure. But you also add this
> > latency to live video streams for example and, depending on the container
> > format, you'll also could violate bitrate constraints because you send the
> > duplicated frames (0 byte theora packets but N bytes container packets) in
> > bursts with the next non-dup frame.
> 
> In this case even changing API in theora will get same problem, it will only
> exclude time machine problem....  (after some thinking) theoraenc create always
> container packets, and if dup detection is working (and it seems to work), it
> create also empthy container packets. So if it is a problem, then it already
> exist.
> 
> I just tested it with mkv, it seems works fine.

But in that case you can send the empty packets between the previous and next
frame instead of sending them after encoding the next frame and then sending
the next frame maybe too late.

> About time machine problem. between two existing frames we have some count of
> missing frames. What happens in the time between this frames, no body know. At
> least there is no evidence.
> How to fill this empty space is up to codec. Fll it with current frame or with
> next frame, it is just assumption about something what we do not know. But we
> need to tell the codec, how many frames are missing.

I guess this starts to become a philosophical discussion now ;) IMHO the
previous frame is always better because it shows what your last observation was
instead of showing the future.

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.


More information about the gstreamer-bugs mailing list