[ivtv-users] Status of the B-Frame issue ?
Levente Novák
lnovak at dragon.unideb.hu
Wed Nov 12 08:23:46 CET 2008
On Mon, 2008-11-10 at 23:00 +0100, Hans Verkuil wrote:
> On Monday 10 November 2008 09:24:44 Alexander Pipelka wrote:
> > Hi,
> >
> > Are there any news on the B-Frame issue.
> > I sent a status log ~ 3 weeks ago.
>
> Hi Alex,
>
> I looked at it today and I could reproduce it. I also found what the
> difference is between the old and current ivtv driver. When you pass
> the B-frames to the card the API says that you should pass B+frames + 1
> to the card. So if there are 0 B-frames, then the number should be 0+1
> = 1. However, for some reason that does not seem to work for a
> cx2584x-based card. If I put in 0 instead of 1, then it works again.
>
> What I don't know is if that really produces a stream without B-frames.
> Applying the attached patch should fix the problem you have, but I
> wonder if you can also test the presence of B/I/P frames and check that
> you get the right results (i.e. really 0 B-frames). I can also test it,
> but not easily and not until after two weeks from now.
>
Tested the patch on a PVR-150 MCE. Here are the observations:
B-frames requested B-frames got Remark
-------------------------------------------------------------
0 1 Plays normally
1 1 Plays normally
2 2 Plays normally
3 3 Choppy
4 4 Choppy and plays at faster speed
So in conclusion: the patch did something but not quite what it was
expected to. At least now with B-frames set to 0 it does save a stream
which however has not 0 but 1 B-frame per P frame. At the higher B-frame
ratios, the video is unusable due to choppiness and/or faster replay
speeds (GOP at 12 frames except at 4 B-frames where it was set to 15 by
qv4l2). The audio seemed normal though.
Levente
More information about the ivtv-users
mailing list