[MPlayer-dev-eng] Re: New inverse-telecine filter

[Andriy N. Gritsenko]

    Hi, D Richard Felker III!

Sometime (on Wednesday, December  3 at 19:57) I've received something...
>On Wed, Dec 03, 2003 at 02:37:50PM +0200, Andriy N. Gritsenko wrote:
>>     Hi, Tristan Seligmann!

>> Sometime (on Tuesday, December  2 at 23:43) I've received something...
>> >Ok, this is a little off topic, but...

>> >On Tue, Dec 02, 2003 at 15:53:47 -0500, D Richard Felker III wrote:
>> >> Something like this is probably a good approach. I didn't write
>> >> anything like this because I'm interested only in maximal accuracy for
>> >> DVD ripping, not TV captures polluted by logos.

>> >Which of the G1 filters would be best for DVD ripping at the moment?

>> I wrote dint filter for the same purpose - to do inverse telecine by
>> deleting extra interlaced frames. It works fine for me at the moment,
>> I've ripped 5 DVDs already with that. :)

>ROTFL!!! This is totally incorrect and will give horrible output. You
>should actually learn what telecine is; then you'd see that you're
>actually throwing away one out of every 4 real frames, so you're
>reducing the framerate from 24 fps to 18 fps in a very choppy way!!

No, from my experience frames (30fps) are:
frame0
frame1
frame2
frame3
frame3+frame4 (interlaced)
frame4

or some alike that. I didn't find a way to get a field instead of frame
and wrote filter just to skip that interlaced frame and it reducing the
framerate from 30 fps to 24 fps well. I've experimented with that and it
finds only 2-5% interlaced frames beside of 24fps framerate so I've
decided it's ok. May be I'm wrong and there is a way for filter to get
not full frame but one field (half of frame) instead so then correct me,
please. :)

    With best wishes.
    Andriy.