[MPlayer-users] Comparison of different deinterlace and denoise filters

D Richard Felker III dalias at aerifal.cx
Wed Oct 22 20:36:58 CEST 2003


On Wed, Oct 22, 2003 at 08:44:31AM +0200, Fabio Papa wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> On Wednesday 22 October 2003 05:38, D Richard Felker III wrote:
> [snip]
> > Once you have interlaced 4:2:0 video like this, you can display it
> > properly (still interlaced) with -vf ilpack. And of course it's ready
> > for deinterlacing or inverse telecine.
> 
> Ok, sorry folks to bump in a second time, but I've similar problems with 
> deinterlacing and I'm interested in all this.
> Rich, you said earlier that scaling operation must be included in the two il 
> in the vf chain, to avoid problems that comes from interlaced videos.
> Then here you're saying that such a video is then ready to be de-interlaced. 
> Hav I got you right? If so, I thought deinterlacing was only good on unscaled 
> videos: am I wrong here?

Yes, definitely wrong. The statement of the rule is "you cannot scale
interlaced video as if the frames are a single picture". If you do so,
you'll ruin it. Of course you're perfectly free to scale the two
fields separately, though. Here's a summary:

2 interlaced fields     --scale-->     trash, output is absolutely
treated as 1 frame                     worthless and ruined

2 interlaced fields     --scale-->     still 2 interlaced fields,
treated as separate                    just scaled to a different
images                                 size

Using the il=d:d and il=i:i filters enables you to (roughly speaking)
treat the two fields independently.

Finally, deinterlacing can be performed on any valid interlaced
frames. The "you can't deinterlace scaled video" myth comes from the
fact that most people stupidly try to scale interlaced video as if it
were a progressive frame, thereby ruining it for ANY further use (not
just for deinterlacing).

Rich




More information about the MPlayer-users mailing list