On Tue, Aug 20, 2019 at 06:13:49PM -0300, James Almer wrote: > On 8/20/2019 3:32 PM, Michael Niedermayer wrote: > > On Tue, Aug 20, 2019 at 02:52:49PM -0300, James Almer wrote: > >> On 8/20/2019 2:46 PM, Michael Niedermayer wrote: > >>> On Tue, Aug 20, 2019 at 02:49:13PM +0200, Paul B Mahol wrote: > >>>> Hi, > >>>> > >>>> I kindly ask that following commit: > >>>> d702769213487923c0fb0abe4b61f4d9ebddb88b > >>>> is reverted as it does contradicts with reference decoder. > >>> > >>> Noone is listed in MAINTAINERs for scpr.c > >>> the main author of scpr.c is you, so i think you are the closest to a > >>> maintainer. > >>> > >>> So if after everyone who wants to comment has commented you still want > >>> this > >>> reverted. Then ill revert (or you can revert too) > >>> but personally i think that skiping duplicated frames has many advantages > >>> and its better to skip them > >>> > >>> Thanks > >> > >> I suggested on IRC to add an option somewhere (new AVCodecContext field? > >> ffmpeg cli option? Mapping it to AV_CODEC_FLAG2_FAST?) to tell crf > >> decoders to instead output vrf in situations like this. > > > > IMHO never output known duplicate frames is best. But if people want to > > change this. A flag/field is the next best option. > > > AV_CODEC_FLAG2_FAST does not solve this as the flag has a different > > meaning. > > How so? The doxy says "Allow non spec compliant speedup tricks.". > Outputting vfr when cfr is the spec compliant behavior sure sounds like > it's a good fit for this feature.
AV_CODEC_FLAG2_FAST will cause output to differ in pixel values from a compliant decoder. while droping duplicated frames can be reversed by duplicating again with no pixel differences. So a user application which needs pixel exact values can use the flag to drop duplicate frames but cannot use AV_CODEC_FLAG2_FAST > > A new flag/field would be needed [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB Elect your leaders based on what they did after the last election, not based on what they say before an election.
signature.asc
Description: PGP signature
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
