On 20 April 2018 at 20:02, Patrick Keroulas < patrick.kerou...@savoirfairelinux.com> wrote:
> From: Damien Riegel <damien.rie...@savoirfairelinux.com> > > This codec is already capable of depacking some combinations of pixel > formats and depth as defined in the RFC4175. The only difference between > progressive and interlace is that either a packet will contain the whole > frame, or only a field of the frame. > > There is no mechanism for interlaced frames reconstruction at the rtp > demux level, so it has to be handled by the codec which needs to > partially recompose an AVFrame with every incoming field AVPacket. > A frame is ouput only when the frame is completed with the 2nd field > (bottom). > > Signed-off-by: Damien Riegel <damien.rie...@savoirfairelinux.com> > Signed-off-by: Patrick Keroulas <patrick.kerou...@savoirfairelinux.com> > --- > > Change v1 -> v2: > Replaced field packets cloning with partial frame decoding. > > @ Rostislav Pehlivanov: Regarding your comment on v1 (thank you for > that), I think we can avoid using the bottom field flag because the > interlaced/progressive format is determined once, at the decoder > initialization, more precisely when the sdp is parsed. RFC4175 > doesn't define any dynamic flag to switch from one to another. > So there is no doubt that 'not top field' means 'bottom field'. > > Doesn't matter what RFC4175 says, the decoder needs to be able to handle any field order sent at any time, including switching to progressive. This isn't a theoretical situation. What if there's packet loss or the user switches streams? What if the user doesn't use the decoder for RFC4175 but rather uses it as a generic unpacker to decode something packed. */ > #define AV_PKT_FLAG_DISPOSABLE 0x0010 > > +/** > + * The packet contains a top field. > + */ > +#define AV_PKT_FLAG_TOP_FIELD 0x0010 > This is wrong, AV_PKT_FLAG_DISPOSABLE == AV_PKT_FLAG_TOP_FIELD. I think there must be a BOTTOM_FIELD flag along with a top field flag, with neither of them flag meaning packet carries progressive. _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
