[MPlayer-users] Re: [-] Question on converting to DivX4...

[Rich Felker]

On Wed, Jul 13, 2005 at 06:15:27AM -0700, RC wrote:
> On Tue, 12 Jul 2005 23:08:51 -0400
> Rich Felker <dalias at aerifal.cx> wrote:
> 
> > ok, you can compute the amount if you like.. i've been too lazy to do
> > it so far, but if you like i can try...
> 
> I would be interested to know, but I base my decision on what looks
> best, not on how close to mathematically perfect it actually is.
> 
> > scale=:240 loses more than half the information with any scaler except
> > sinc. 
> 
> Okay, even with -sws 8, pp=lb is still clearly better.

sinc looks very bad. :)
but it's the only near-lossless scaler (i.e. upscaling the scaling
back down with sinc should give results almost identical to the
original).

> > see, when you blend 2 fields together, the information lost is not
> > the details of the actual fields (altho some of that is lost too),
> > but rather which information goes with which field. you've reduced
> > 60fps to 30fps in an essentially irreparable way..
> 
> Well, I never claimed it could be easily reversed.  Are there any
> deinterlacers (other than tfields) that can be easily reversed?  I
> doubt it.

any correct deinterlacer (that outputs 60fps) is almost entirely
reversible. naturally anything that destroys have the framerate is not
going to be reversible.

btw it's possible to make a 60fps version of pp=lb, and it would
probably look very good to you. naturally it has no flicker, only
ghosts. however, i still don't think it would be reversible, for the
same reason pp=lb isn't reversible on a single frame (only now the
argument is in the time dimension).

rich