[MEncoder-users] Product suggestion: Encode Decision List

James Hastings-Trew jimht at shaw.ca
Mon Oct 8 02:34:19 CEST 2007


RC wrote:
>> Here's one background example. If a movie contains both progressive as
>> well as interlaced content, you generally have three choices:
>>
>> 1) encode as if it's all progressive if you don't worry too much about
>> best representation of interlaced content
>> 2) encode as if it's all interlaced if you don't worry too much about
>> best representation of progressive content
>> 3) if you are concerned about best quality, split the file at each
>> content type boundary, encode each segment separately and then
>> recombine
>>     
>
> A good deinterlacer, like yadif and pp=lb, shouldn't have noticable
> (negative) effects on progressive video, so I don't see this as much of
> an issue.
>
> The same is true for some ivtc filters (namely: pullup).
>   
I agree with RC. My methodology is to detect the frame rate of the 
source (using MPlayer) - if its film rate I use the pullup,softskip 
filter. If its mixed rate and mostly film or hard telecine, I use 
pullup,softskip,yadif=0. If its video rate I just use pp=lb. yadif=0 
won't do much of anything to a progressive frame. No matter what you do 
with video, deinterlacing is going to leave some kind of artifact - 
either as ghosts, or motion hiccups - because you are altering the 
contents of each frame. I prefer to have smooth motion so I use the 
simpler filter (pp=lb) on such material.

WMV is a format that allows for variable frame rate, and the problem is 
that once content is encoded that way, it becomes very problematic to 
convert it to any other format - you have to "discover" the original 
frame rate or pick a very high frame rate that is common to both film 
and video (119.88 fps), and duplicate the missing frames. Either option 
is a pain in the ass. The point I am trying to make here is that 
variable frame rate formats are not very common, or portable, and in my 
opinion should be avoided accordingly. Just putting that out there.



More information about the MEncoder-users mailing list