| Message ID | 1472140695-10290-1-git-send-email-derek.buitenhuis@gmail.com |
|---|---|
| State | Accepted |
| Commit | 04f8d312877ffdcb816c7ff74b94eaa06dd6e1f0 |
| Headers | show |
On Thu, Aug 25, 2016 at 04:58:14PM +0100, Derek Buitenhuis wrote: > This breaks files with legitimate single-entry edit lists, > and the hack, introduced in f03a081df09f9c4798a17d7e24446ed47924b11b, > has no link to any known sample in its commit message. > > Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com> > --- > libavformat/isom.h | 1 - > libavformat/mov.c | 8 +------- > 2 files changed, 1 insertion(+), 8 deletions(-) ok if i push with the changes requested in: https://lists.ffmpeg.org/pipermail/ffmpeg-devel/2016-August/198303.html ? (seems another reply with forgotten CC) Thanks [...]
On 8/25/2016 5:18 PM, Michael Niedermayer wrote: > ok if i push with the changes requested in: > https://lists.ffmpeg.org/pipermail/ffmpeg-devel/2016-August/198303.html > ? (seems another reply with forgotten CC) You can change 'hack' to 'incorrect workaround' if you wish. However, removing the part about the sample being missing in the commit message would be wrong, because it is, in fact, not linked in the commit message. - Derek
On 8/25/2016 5:49 PM, Derek Buitenhuis wrote: > However, removing the part about the sample being missing in > the commit message would be wrong, because it is, in fact, not > linked in the commit message. I must step out now, so please push whenever some consensus is reached. No point in bikeshedding over a commit message, and whilst I think it'd be silly to remove this bit, I do not feel strongly about it. - Derek
diff --git a/libavformat/isom.h b/libavformat/isom.h index df6c15a..2246fed 100644 --- a/libavformat/isom.h +++ b/libavformat/isom.h @@ -155,7 +155,6 @@ typedef struct MOVStreamContext { MOVDref *drefs; int dref_id; int timecode_track; - int wrong_dts; ///< dts are wrong due to huge ctts offset (iMovie files) int width; ///< tkhd width int height; ///< tkhd height int dts_shift; ///< dts shift when ctts is negative diff --git a/libavformat/mov.c b/libavformat/mov.c index 1bc3800..54c63ad 100644 --- a/libavformat/mov.c +++ b/libavformat/mov.c @@ -2802,12 +2802,8 @@ static void mov_build_index(MOVContext *mov, AVStream *st) sc->time_offset = start_time - empty_duration; current_dts = -sc->time_offset; if (sc->ctts_count>0 && sc->stts_count>0 && - sc->ctts_data[0].duration / FFMAX(sc->stts_data[0].duration, 1) > 16) { - /* more than 16 frames delay, dts are likely wrong - this happens with files created by iMovie */ - sc->wrong_dts = 1; + sc->ctts_data[0].duration / FFMAX(sc->stts_data[0].duration, 1) > 16) st->codecpar->video_delay = 1; - } } if (!unsupported && st->codecpar->codec_id == AV_CODEC_ID_AAC && start_time > 0) @@ -5352,8 +5348,6 @@ static int mov_read_packet(AVFormatContext *s, AVPacket *pkt) sc->ctts_index++; sc->ctts_sample = 0; } - if (sc->wrong_dts) - pkt->dts = AV_NOPTS_VALUE; } else { int64_t next_dts = (sc->current_sample < st->nb_index_entries) ? st->index_entries[sc->current_sample].timestamp : st->duration;
This breaks files with legitimate single-entry edit lists, and the hack, introduced in f03a081df09f9c4798a17d7e24446ed47924b11b, has no link to any known sample in its commit message. Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com> --- libavformat/isom.h | 1 - libavformat/mov.c | 8 +------- 2 files changed, 1 insertion(+), 8 deletions(-)