[mythtv] [PATCH] Smoother "Experimental A/V Sync"

[James Armstrong]

Doug Larrick wrote:

> The attached small but important patch fixes a longstanding bug with 
> the ExAVSync method, making it produce a much smoother frame cadence.
>
> Before, if the display time drifted too close to the beginning of the 
> frame interval, the code would pull the frame into the previous frame 
> interval, causing a skipped frame (I believe this was a bug, not 
> intentional).  Since this frame was displayed too early, the following 
> frame would be displayed for too many frame intervals to try to catch 
> up, producing a very noticeable glitch.  In addition, there was no 
> code to prevent starting the display of a particular frame very late 
> in the frame interval.  I believe that when displaying HDTV-sized 
> images, this could leave insufficient time for the video card to 
> display the frame, resulting in periods of significant video glitching.
>
> The new code nudges the display time backwards or forwards to keep it 
> in a reasonable position within the frame interval, which I believe 
> was the original intent of the existing code.
>
> The patch also decouples "Jitter Reduction" from "Experimental A/V 
> Sync".  Before, it was necessary to select both of these in order to 
> use the videocard-based sync method.
>
> I've also taken the liberty to rename "Experimental A/V Sync" in the 
> UI to the more-descriptive "Video card A/V Sync", and updated the help 
> text to mention the two hardware methods supported (nVidia and OpenGL).
>
> If you'd like to see some timing data I collected after making this 
> change, take a look at 
> http://jekyl.ddts.net/doug/myth-log-delay-old.gnumeric (Gnumeric 
> spreadsheet).
>
> -Doug

Would this explain a problem I have when watching some recordings? I 
notice it alot on American Chopper, every now and then I see what looks 
like a flash on the screen of a different video frame, that goes away 
and the recording plays as normal.

- James