[MPlayer-dev-eng] [PATCH] deinterlace property fix

[Vicente Sendra]

--- El sáb, 20/10/12, Andy Furniss wrote:

> De: Andy Furniss <andyqos at ukfsn.org>
> Asunto: Re: [MPlayer-dev-eng] [PATCH] deinterlace property fix
> Para: mplayer-dev-eng at mplayerhq.hu
> CC: "Vicente Sendra" <visenri at yahoo.es>
> Fecha: sábado, 20 de octubre, 2012 22:33
> Vicente Sendra wrote:
> > I've previously send a fix for command.c in:
> > Better double frame rate output and framedrop
> > 
> > But there was a bug in my code.
> > 
> > This new one fixes old problem:
> > 
> > Better handling of deinterlace property, returns
> M_PROPERTY_ERROR if it can't be changed (or read), and
> displays osd value correctly (when change was not successful
> it showed the opposite value).
> > 
> > And my bug:
> > 
> > if VFCTRL_GET_DEINTERLACE was not successfull,
> deinterlace value was undefined.
> > 
> > I send it as a separate fix, as it's unrelated to new
> framedrop system.
> 
> I am interested in your work as I have been trying to get
> field rate = refresh to work recently (I am sure long ago it
> worked better than now, but this may be radeon/mesa related
> rather than mplayer - or perhaps my test cases were
> easier).
> 
> Testing your framedrop patches + this one I see a problem
> with -vo gl (I am not deinterlacing).
> 
> A 50fps progressive HD file speeded up to 60fps to match
> refresh works with vanilla mplayer, but drops lots of frames
> with your patches.
> 
> vblank_mode=2 mplayer -cache 20000 -demuxer lavf -speed
> 30/25 -fs -lavdopts threads=4 -vo gl -framedrop samp50p.mkv
> 
> One other thing - to deinterlace HD although I have 4x
> 3.4Ghz CPUs I only really expect to be able to use tfields=0
> or preferably -vo vdpau:deint=2 (radeon not nvidia - not
> available without mesa hack) so if you plan on continuing
> this good work maybe vdpau code needs a patch as well.

Is it dropping frames if you play at it's normal speed (50fps)?

What screen refresh rate are you using?, if you are using 60fps (and playing at 60fps), my new frame drop system drops frames every now and then (on my system, 1-2 frames every 2-3 seconds), this is normal, because mplayer timing is not exactly equal to the screen refresh rate, and we have to drop some frames to compensate for this little difference.

New frame drop system assumes that page flip should take almost 0 time, if page flip takes too much compared to total frame time, it stats dropping frames.

Can i have access to your samples?