github kajisho5/ffmpeg-skill v3.8.3

2 hours ago

Automated release: version and notes generated from pull requests merged since 3.8.2.

Fixed

  • overlay.py, waveform.py, loop.py and sync.py --replace-audio end their sound to the
    sample; pad.py and freeze.py add silence to the sample
    (the siblings of #493). The first
    four ended their output with -t {:.3f}, a length to the millisecond: an input whose sound is
    59244 samples long (48 kHz, 1.23425 s) came out of overlay.py --image/--video and
    waveform.py with 59232 (54433 at 44.1 kHz with 54419), the reference's 59244 came out of
    sync.py --replace-audio as 59232, and loop.py --duration 2.23425 wrote 107232 for 107244. Where
    the sound runs on past the length (sync.py's pad, loop.py's next repeat) a millisecond that
    rounds up wrote too much: 59280 for a reference of 59261. pad.py --start/--end and
    freeze.py --hold (a hold at the clip's end or at its first frame) wrote the silence as
    adelay=<whole ms> / apad=pad_dur={:.3f}: --end 0.23425 added 11232 samples for 11244, and
    the millisecond was cut, not rounded, so --start 1.001 delayed the sound by 1.000 s
    (int(1.001 * 1000) is 1000; 741 of the whole-millisecond values up to 100 s). The -t is the
    length to the microsecond when the millisecond text names another sample count
    (length_output_t() in _common), and a silence that is not a whole millisecond is counted in
    samples (silence_filter(): adelay=NS, apad=pad_len=N behind an aformat that pins the
    rate). A length or a silence in whole milliseconds, an input without sound and a dry run's
    pending input write the command they always did, except the whole milliseconds that were cut by
    one. The picture is bounded by the same -t, so the limit carries no half sample of slack
    (with it a layer's frame that starts exactly at the main input's end came through on a 1/60000
    time base); a frame that starts in the last half millisecond before the end, which -t {:.3f}
    left out when it rounded down, is now written. pad.py's and freeze.py's picture
    (tpad, whole frames) is unchanged. Measured on FFmpeg 6.1.1.
  • fix: overlay, waveform, loop and sync --replace-audio end their sound to the sample; pad and freeze add silence to the sample (#499)

What's Changed

  • fix: overlay, waveform, loop and sync --replace-audio end their sound to the sample; pad and freeze add silence to the sample by @kajisho5 in #499

Full Changelog: v3.8.2...v3.8.3

Don't miss a new ffmpeg-skill release

NewReleases is sending notifications on new releases.