Problems with resync in libxine when using vdpau

Ralph Scharpf <[email protected]> Sat, 15 Oct 2011 12:42:21 +0200
Newsgroups gmane.comp.video.xine.user
Message-ID <[email protected]>
I am using yavdr on an Atom ION and vdr-sxfe for viewing full hd video. vdr-
sxfe is based on libxine 1.2.0. For decoding libvdpau is used. 

When viewing full hd from german ARD or ZDF in format 720p/50Hz my vdr 
delivers sometimes wrong frames. These seems to come from my tuner. I will 
investigate this in parallel. 

The problem with libxine is: After the error in stream is over there remain 
artefacts in the video stream. I recognize them mainly at the shape of 
objects. Not all frames have artefacts but more than 2 per second.

I was able to create a short recoding that also shows the error. Any libxine 
based player I use for viewing the video shows the artefacts. When I use 
"mplayer -vc ffh264vdpau" (sorry, I don't want to create any pressure) I  also 
see the error in steam but after the stream is ok again there remain no 
artefacts in the video. On an other machine with a strong PhenomII CPU but no 
vdpau the resync is successful with libxine, too.

I would like to ask if a xine developer would be so kind to have a look at 
the recoding with the problem on a NVidia/vdpau HW. I am sure that there is a 
way to resynchonze the stream without the remaing artefacts.

I had several updates of related packets in last months but the problem was 
present all the time. It seems not to be a problem related to a specific 
release. The version string from my current libxine version is 
1.2.0~hg20111014.0440-0yavdr0~natty. vdpau is at version 0.4.1-2. The nvidia 
driver has version 285.05.09-0. 

I also tried to tune parameters in xine configuration. But the problem can not 
be solved by tuning parameters. The compositing is completely disabled 
(xorg.conf) on the system.


------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2d-oct