[FFmpeg-devel] [PATCH] lavfi/mp: remove mp=dint

Paul B Mahol onemda at gmail.com
Thu Sep 5 17:00:13 CEST 2013


On 9/5/13, Stefano Sabatini <stefasab at gmail.com> wrote:
> On date Wednesday 2013-09-04 21:19:33 +0200, Reimar Doeffinger encoded:
>> On Wed, Sep 04, 2013 at 11:50:47AM +0000, Paul B Mahol wrote:
>> > Signed-off-by: Paul B Mahol <onemda at gmail.com>
>> > ---
>> >
>> > I can't comprehend its documentation.
>> > Code is extremly unreadable so I can't bother what it actually does.
>> > There is better filter that can detect interlaced (among others)
>> > frames.
>> > There is better filter that can drop any (including interlaced) frames.
>>
>
>> What it does is quite simple: It detects frames that look (still)
>> interlaced and tries to drop them.
>> I expect is intended to be used with incorrectly inverse telecined
>> content
>> to get rid of any bad-looking frames.
>> Due to this, it will never drop more than one frame in a row, since
>> that would look bad and is likely to mean that the interlace detection
>> didn't work right.
>> I don't know if the two filters you mentioned can achieve this
>> behaviour together, though I would guess it should not be hard to
>> implement if they do not.
>
> Theoretically, you could combine idet and select. Practically this
> requires at least two steps, mark interlaced frames as such through
> metadata, and allow the select filter to access metadata
> (e.g. select=if(metadata(lavfi.idet.interlaced))).
>
> As alternative, scripting would allow that easily with user-provided
> filtering logic.
>
> I also wonder how the idet detection code compares with the one in
> dint (we should probably factorize and pick the best one).

Is anybody using this filter?


More information about the ffmpeg-devel mailing list