[FFmpeg-user] alternating checkerboard

Paul B Mahol onemda at gmail.com
Tue Feb 9 12:59:20 EET 2021


On Tue, Feb 9, 2021 at 10:02 AM Paul B Mahol <onemda at gmail.com> wrote:

>
>
> On Tue, Feb 9, 2021 at 9:45 AM Mark Filipak (ffmpeg) <markfilipak at bog.us>
> wrote:
>
>> On 02/09/2021 03:14 AM, Michael Koch wrote:
>> > Am 09.02.2021 um 04:49 schrieb Mark Filipak (ffmpeg):
>> >> I don't know the details of the filters that work on pixels ('blend',
>> for example). I hope a
>> >> knowledgeable person will respond with suggestions.
>> >>
>> >> I'd like to experiment with a filter that mixes pixels from 2
>> progressive streams on an
>> >> alternating checkerboard pattern. Got any suggestions?
>> >>  _________________    _________________    _________________
>> >> |# # # # # # # # #   | % % % % % % % %    |#%#%#%#%#%#%#%#%#
>> >> | # # # # # # # #    |% % % % % % % % %   |%#%#%#%#%#%#%#%#%
>> >> |# # # # # # # # # + | % % % % % % % %  = |#%#%#%#%#%#%#%#%#
>> >> | # # # # # # # #    |% % % % % % % % %   |%#%#%#%#%#%#%#%#%
>> >> |# # # # # # # # #   | % % % % % % % %    |#%#%#%#%#%#%#%#%#
>> >>
>> >>
>> >
>> > Have a look at the "maskedmerge" filter. In your case the mergemap file
>> would contain a pattern of
>> > 0's and 255's. This file can be generated with "geq" filter (see the
>> example in chapter 2.107).
>>
>> Thanks for the suggestion. Actually, it appears that the 'blend' filter
>> does what I want, but I'm
>> still testing to be sure.
>> blend=all_expr='if(eq(mod(X,2),mod(Y,2)),TOP,BOTTOM)'
>> 'blend' appears to input 2 frames in 2 streams and output 1 frame (with
>> unknown PTS). 'tblend', on
>> the other hand, appears to take 2 frames in 1 stream and output 2 frames,
>> not 1 frame, (again, with
>> unknown PTS). It sure would be nice if these filter functions were
>> adequately documented, especially
>> regarding how they determine output PTS. But you shouldn't listen to me
>> because, to quote the
>> authorities: "That makes no sense".
>>
>>
> Why you write unknown PTS? What you are talking about, please do not be
> dense.
>
> PTS is always deterministic from these filters.
>

ffmpeg -i ~/Videos/red-leaf-tips.avi -i ~/Videos/musical-box.avi -lavfi
format=yuv444p,framepack=format=frameseq,stereo3d=in=al:out=chl,setsar=1/2
-f nut -c:v rawvideo -|mpv -

It is not exactly what you want, as this variant do not discard data. But
you get an idea.

>


More information about the ffmpeg-user mailing list