[mythtv-users] Bob deinterlacing drops frames, but not high CPU

[Matt Doran]

Mark Kendall wrote:
> On 5/16/07, Matt Doran <matt.doran at papercut.biz> wrote:
>   
>> Doug Larrick wrote:
>> It's 25fps DVB broadcast (I'm in australia).  I've got my X vertical
>> refresh is 50Hz, because I was under the impression this should lead to
>> smoother scrolling (i.e. an integer multiple of the frame rate).
>>
>> Is that what you were after?
>>     
>
> I've had a similar problem before where the frontend was detecting the
> screen refresh rate as something like 49.97 (rather than 50) and
> deciding that the display couldn't cope with bob.
>
> To confirm, you'll need to check your frontend logs (-v playback).
>
> If true, the 'easiest' way to fix it is to slightly modify your
> modeline so that the timings calculated by X/mythtv are give a refresh
> rate >50Hz.
>
>
>   
Thanks Mark, but I think bob is actually working.  I had -v playback 
enabled, and here's some output.

    2007-05-16 20:54:55.567 AFD: Successfully opened decoder for file:
    "/var/lib/mythtv/2512_20070516205454.mpg". novideo(0)
    2007-05-16 20:54:55.570 NVP: DoPlay: rate: 25 speed: 1 skip: 1 =>
    new interval 40000
    2007-05-16 20:54:55.571 Avg read interval was 199 msec. 64K block size
    2007-05-16 20:54:55.573 Set video sync frame interval to 40000
    2007-05-16 20:54:55.574 Set video sync frame interval to 40000
    2007-05-16 20:54:55.576 Using deinterlace method bobdeint
    2007-05-16 20:54:55.576 NVP: Stretch Factor 1, allow passthru
    2007-05-16 20:54:55.576
    RingBuf(/var/lib/mythtv/2512_20070516205454.mpg):
    CalcReadAheadThresh(2695883664 KB)
                             -> threshhold(64 KB) min read(0 KB) blk
    size(32 KB)
    2007-05-16 20:54:55.576 Resyncing position map. posmapStarted = 0
    livetv(1) watchingRec(1)
    2007-05-16 20:54:55.577 Position map filled from DB to: 0
    2007-05-16 20:54:55.577 SyncPositionMap watchingrecording, from DB:
    1 entries
    2007-05-16 20:54:55.578 Filling position map from 1 to 17
    2007-05-16 20:54:55.578 Position map filled from Encoder to: 0
    2007-05-16 20:54:55.578 SyncPositionMap watchingrecording total: 1
    entries
    2007-05-16 20:54:55.587 Avg read interval was 196 msec. 64K block size
    2007-05-16 20:54:55.681 NVP: Waiting for prebuffer.. 2
    AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
    2007-05-16 20:54:55.847 NVP: Waiting for prebuffer.. 3
    AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
    2007-05-16 20:54:55.967 Avg read interval was 185 msec. 96K block size
    2007-05-16 20:54:56.011 NVP: Waiting for prebuffer.. 4
    LAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
    2007-05-16 20:54:56.175 NVP: Waiting for prebuffer.. 5
    LAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
    2007-05-16 20:54:56.339 NVP: Waiting for prebuffer.. 6
    UuUULAAAAAAAAAAAAAAAAAAAAAAAAAA
    2007-05-16 20:54:56.503 NVP: Waiting for prebuffer.. 7
    UuUULAAAAAAAAAAAAAAAAAAAAAAAAAA
    2007-05-16 20:54:56.667 NVP: Waiting for prebuffer.. 8
    UUUUUUUuUULAAAAAAAAAAAAAAAAAAAA
    2007-05-16 20:54:56.691 Avg read interval was 193 msec. 128K block size
    2007-05-16 20:54:56.831 NVP: Waiting for prebuffer.. 9
    UUUUUUUuUULUuAAAAAAAAAAAAAAAAAA
    2007-05-16 20:54:57.635 Avg read interval was 192 msec. 160K block size
    2007-05-16 20:54:57.830 NVP: Video is 5.53125 frames behind audio
    (too slow), dropping frame to catch up.
    2007-05-16 20:54:57.835 NVP: Video is 9.36092 frames behind audio
    (too slow), dropping frame to catch up.
    2007-05-16 20:54:57.839 NVP: Video is 12.0144 frames behind audio
    (too slow), dropping frame to catch up.


Another thing that may or may not be useful.  I'm not using the OpenGL 
sync method, because when I enable it, I get a very subtle mismatch 
between audio and video ... and no frame drop messages.  So I'm using 
the RTC video timing.  

Thanks,
Matt