diff options
| author | Carl Hetherington <cth@carlh.net> | 2022-01-20 21:23:09 +0100 |
|---|---|---|
| committer | Carl Hetherington <cth@carlh.net> | 2022-01-24 16:01:25 +0100 |
| commit | 86c9cd50e1b73e6d89db312f2ee58e85ff270194 (patch) | |
| tree | 40cdd26c7a428b140645b5d42d4119fc98dc4f04 /src/lib/player.cc | |
| parent | 645936f18657c8d5c275cbf3443974784304d672 (diff) | |
Slightly hacky fix for -114 errors during 3D encodes (#2173).v2.15.188
It still feels like I'm missing some large and beautiful trick
with gathering all the gap-filling code into one place, but
such a trick still eludes me.
Diffstat (limited to 'src/lib/player.cc')
| -rw-r--r-- | src/lib/player.cc | 13 |
1 files changed, 13 insertions, 0 deletions
diff --git a/src/lib/player.cc b/src/lib/player.cc index 4e586ed77..42c22ab7d 100644 --- a/src/lib/player.cc +++ b/src/lib/player.cc @@ -812,6 +812,19 @@ Player::pass () for (auto const& i: _delay) { do_emit_video(i.first, i.second); } + + /* Perhaps we should have Empty entries for both eyes in the 3D case (somehow). + * However, if we have L and R video files, and one is shorter than the other, + * the fill code in ::video mostly takes care of filling in the gaps. + * However, since it fills at the point when it knows there is more video coming + * at time t (so it should fill any gap up to t) it can't do anything right at the + * end. This is particularly bad news if the last frame emitted is a LEFT + * eye, as the MXF writer will complain about the 3D sequence being wrong. + * Here's a hack to workaround that particular case. + */ + if (_next_video_eyes && _next_video_time && *_next_video_eyes == Eyes::RIGHT) { + do_emit_video (black_player_video_frame(Eyes::RIGHT), *_next_video_time); + } } return done; |
