[MPlayer-users] Re: Crop before deinterlacing OK?

[Matthias Wieser]

Am Montag, 25. Juli 2005 00:48 schrieb Rich Felker:
> On Sun, Jul 24, 2005 at 07:31:13PM +0200, Matthias Wieser wrote:
> > Am Sonntag, 24. Juli 2005 04:40 schrieb RC:
> > > On Sat, 23 Jul 2005 18:12:54 +0200
> > >
> > > Matthias Wieser <mwieser at gmx.de> wrote:
> > > > Maybe I am the only one who does not like pp=lb
> > > > because of this effect:
> > >
> > > No, you're definately in the majority.  I seem to be almost the
> > > only person around here using pp=lb.
> >
> > Have you ever tried pp=l5? It seems to be a variant of pp=lb which
> > produces slightly more sharpness. I think it should not have any
> > disadvantages.
>
> It has all the same disadvantages as lb, plus the disadvantages of a
> sharpener (hurting compressibility and not improving quality).

Maybe not improving psnr but I think it improves visual quality. I have 
done some tests with different test patterns and the deinterlaced result 
of pp=l5 is much more comparable to the original as the one from pp=lb.

(btw.: when measuring the psnr between the original and the scaled down 
and encoded video, sws=2 or sws=6 often results in higher PSNR than sws=1 
or sws=7)

> It's 
> exactly the same thing as running pp=lb followed by a simple -1,4,-1
> sharpening filter.

But as pp=lb by the very nature of the system introduces some vertical 
blurring this sharpening might improve the overall quality.

> > > The ghosting looks dramatic in still images, but it doesn't bother
> > > me much at 30fps (except VERY fast scrolling text, which is rare).
> >
> > With TFTs it is getting even worse because pp=lb lowers the contrast
> > at the edges of moving objects which rises the response time of TFTs.
>
> I don't think this has anything to do with it..

The response time of most TFTs is dependent on the contrast between the 
two colours. Sharp edges as they are produced by pp=fd result in higher 
local contrast between the actual and the last frame which in result is 
good for TFTs.
For example: a black object is moving over white ground.
when using pp=fd the response time from black to white and from white to 
black is relevant. 
When using pp=lb the respons timee from black to gray, gray to white, 
white to gray and gray to black are relevant which often need much more 
time than switches between black and white.

Regards,
  Matthias