[FFmpeg-user] Avoid yadif auto format for progressive bgra

Christian Ebert blacktrash at gmx.net
Tue May 8 10:02:35 CEST 2012


* Tim Nicholson on Tuesday, May 08, 2012 at 08:16:47 +0100
> On 07/05/12 07:55, Robert Nagy wrote:
>> I have a problem with yadif where a progressive bgra video stream is
>> converted to yadif supported formats even though yadif does not do anything
>> to the stream (auto_enable==true).
> 
> I would hazard a guess that in order for YADIF to decide to do nothing
> it has to be able to understand the stream first.
> 
>> My question is whether I can solve this using a split and some form of
>> "conditional" filter? e.g.
>> 
>>                      /-> if interlaced   -> auto-format -> yadif ->\
>> source->split ->
>> -> merge -> sink
>>                      \-> if progressive ->                              ->/
>> 
>> Is this possible? And will it work as I assume it would? I guess it would
>> not work for streams with both interlaced and progressive frames, but does
>> such stream exist?
> 
> In terms of flagging hopefully not. In terms of content, often. Which
> begs the question, how do you determine the whether or not material is
> interlaced or not?
> 
> The idet filter makes a good stab at content based analysis *but* it
> doesn't work on bgra sources as it auto inserts a scale 0 filter to
> convert to YUV first and any auto inserted filter seems to stuff its
> analysis.
> 
> Or do you trust the interlace flag. this requires that material is both
> flagged correctly by the coding system, *and* the flag is correctly
> interpreted by ffmpeg. Currently results suggest that this is not nearly
> good enough to be relied upon.

For my inputs it seems the interlace flag is good enough, I query
it with:

mediainfo --Inform="Video;%ScanType%"

Is there a similar command for ffprobe?

c
-- 
  Was heißt hier Dogma, ich bin Underdogma!
[ What the hell do you mean dogma, I am underdogma. ]
free movies       --->>> http://www.blacktrash.org/underdogma
http://itunes.apple.com/podcast/underdogma-movies/id363423596


More information about the ffmpeg-user mailing list