[mythtv] 2019-devel: lipsync drift
johnpilk222 at gmail.com
Tue Nov 26 10:43:27 UTC 2019
On 26/11/2019 08:56, Mark Kendall wrote:
> Apologies for the delay - I've been away with the fairies for a few days:)
> Can you please post full logs for the working and non-working cases?
> The snippets are useful but I need the 101 other pointers that the
> logs might give.
> Given the nature of the problem, I'd suggest with '-v playback,gpu
> --loglevel=debug' logging.
> If you want to go 'hardcore' as well then add timestamp and audio
> options - though the logs will be fairly lengthy:)
> thanks and regards
> On Mon, 25 Nov 2019 at 23:57, John Pilkington <johnpilk222 at gmail.com> wrote:
>> On 24/11/2019 22:41, John Pilkington wrote:
>>> To continue...
>>> I have a 'new' F30 box, i3 4-core with nVidia and a single VGA-driven
>>> monitor, again running c794391ad. The only recordings at present are
>>> from assorted iptv channels, but they show no obvious a/v drift.
>>> The DispRes: Desktop video mode on frontend startup and the DispRes:
>>> SwitchToGUI on exit are both clean 1920x1080 60.000 Hz
>> I commented earlier on the size of the timing StdDev shown in the
>> terminal log when the Playback Data OSD is active. It was around 25 ms,
>> with a frame interval of 40 ms. The SD appears to track the correction
>> increment set for AVSync2, which suggests that a correction is being
>> applied on every frame. I had 30 ms, and now have 2 ms. There is still
>> a drift, though.
The recording has been through my usual demux by Project-x and remux by
mplex process, which has delivered good sync over seveal years. It has
speech to camera and shows the problem.
AVSync2 2 ms, Standard decode, 2 cpus, OpenGL, HQ deinterlace, no ticks
mythfrontend -v playback,gpu --loglevel=debug | tee lipsync.log
More information about the mythtv-dev