[MPlayer-dev-eng] [PATCH] synch to new x264 revision

[Guillaume POIRIER]

Hi,
Le ven 24/09/2004 à 19:19, Loren Merritt a écrit :
> >
> > fullinter improves PSNR 40% of the time
> 
> ok. Does it produce a net improvement when you include the cases where it 
> doesn't help? 

I didn't save PSNR stats, so I can't tell you right now. I'm gonna test
that when I have some more time.


> Also, fullinter's performance is probably correlated with 
> subq. Do you have any stats on when it helps?

Ok, I note that I should test too.


> > deblock helps 50% of times
> >
> > deblockalpha doesn't help 73% of the time,
> > helps when set to 1 13% of the times
> >
> > deblockbeta doesn't help 73% of times
> 
> deblock isn't about PSNR (though deblock=true,deblockalpha=0,deblockbeta=0 
> is usually the best PSNR in my tests). It's about what you think is the 
> best tradeoff between detail and blocking.

That's the reason why this needs to be documented, doesn't it? ;-)
Well, anyway, with a little bit of good sense, I guess people wouldn't
bother writing those kind of options it they were always bad! ;-)


> > iframe interval does need to be increased. If helps when set higher than
> > 100 50% of time
> 
> And iframe>100 hurts the other 50% or the time? Or did scenechange 
> detection kick in and not actually use >100 iframe interval?

iframe tests I made were kinda irrelevant. It tested (60, 80, 90, 100,
100, 120, 130) keyframe intervals. When I'll re-do my test, I'll try to
look for scene detection.


> > qcomp helps when set to 0.6 50% of times
> 
> qcomp=0.6 is the default. What other values is this compared to?

(0.4, 0.5, 0.6, 0.7, 0.8). Isn't it great that default is _that_ good?
;-)


> > subq helps when set to 5 or 3 50% of times
> > and helps when set to 0 33% of times
> 
> really? subq=0 should _always_ be worse than subq=1 by a large margin,
> (0 searches with only halfpel precision, but still takes the bits to 
> store qpel) and subq>1 should _always_ be better, though the amount of 
> difference depends on the source video.

I also though it was weird. But I thought it was maybe like qpel in
"regular" MPEG-4 that doesn't help PSNR but improves the
precision/details of the picture.
Is that the same?


> > iframes help when set to 2 47% of times, when set to 3 33% of times
> 
> what option is "iframes"? do you mean "idrframe"?

Yeah, That's what I meant. Sorry.
That confirms that anime seems to like idrframes.


Regards,
Guillaume