diff options
author | Anton Khirnov <anton@khirnov.net> | 2023-05-28 16:22:58 +0200 |
---|---|---|
committer | Anton Khirnov <anton@khirnov.net> | 2023-06-05 16:15:04 +0200 |
commit | ccf219e361957eef8fab45ce8b02cdf4d42e5e28 (patch) | |
tree | 046bc81e14869a524392bc7eecca2702b857e082 /fftools/ffmpeg_enc.c | |
parent | c7a05ac117bc4152b27aa6127d67cb5bc0f09271 (diff) |
fftools/ffmpeg_enc: avoid breaking exactly integer timestamps in vsync code
The code will currently add a small offset to avoid exact midpoints, but
this can cause inexact results when a float timestamp is exactly
representable as an integer.
Fixes off-by-one in the first frame duration in multiple FATE tests.
Diffstat (limited to 'fftools/ffmpeg_enc.c')
-rw-r--r-- | fftools/ffmpeg_enc.c | 6 |
1 files changed, 4 insertions, 2 deletions
diff --git a/fftools/ffmpeg_enc.c b/fftools/ffmpeg_enc.c index fab71ca0c0..d390e384ed 100644 --- a/fftools/ffmpeg_enc.c +++ b/fftools/ffmpeg_enc.c @@ -888,9 +888,11 @@ static double adjust_frame_pts_to_encoder_tb(OutputFile *of, OutputStream *ost, float_pts = av_rescale_q(frame->pts, filter_tb, tb) - av_rescale_q(start_time, AV_TIME_BASE_Q, tb); float_pts /= 1 << extra_bits; - // avoid exact midoints to reduce the chance of rounding differences, this + // when float_pts is not exactly an integer, + // avoid exact midpoints to reduce the chance of rounding differences, this // can be removed in case the fps code is changed to work with integers - float_pts += FFSIGN(float_pts) * 1.0 / (1<<17); + if (float_pts != llrint(float_pts)) + float_pts += FFSIGN(float_pts) * 1.0 / (1<<17); frame->pts = av_rescale_q(frame->pts, filter_tb, enc->time_base) - av_rescale_q(start_time, AV_TIME_BASE_Q, enc->time_base); |