Notes on ffmpeg

Transcode audio with metadata
Encode/compress raw video and audio
Screencast on Linux with PulseAudio
- With microphone
Video to GIF
- Online
Trim/cut video to video
Misc
Videos with wrong orientation in metadata
Map video and audio streams into single output
References

Transcode audio with metadata

ffmpeg and quodlibet/operon

	for src in *flac; do
	dst="$(echo $src \| sed -e 's/\.flac//' -e 's/ /_/g')".mp3
	ffmpeg -i "$src" -ab 320k "$dst"
	operon clear -a "$dst"
	operon copy "$src" "$dst"
	done

view raw ffmpeg_quodlibet_operon.sh hosted with ❤ by GitHub

Encode/compress raw video and audio

Compress raw video into Matroska container, see Digitizing Hi8 cassettes & Digitizing VHS cassettes.

	date >> log.txt
	echo 'Starting job' >> log.txt
	sudo systemctl stop fahclient.service
	sudo systemctl status --full fahclient.service

	for src in *avi ; do
	date >> log.txt ;
	echo "Starting conversion for $src" >> log.txt ;
	dst="$(echo $src \| sed -e 's/\.avi$//')".mkv ;
	ffmpeg -ac 2 -channel_layout stereo -i "$src" -c:v libx264 -preset slow -crf 17 -pix_fmt yuv422p -c:a pcm_s16le "$dst" ;
	date >> log.txt ;
	echo "Finished conversion, created $dst" >> log.txt ;
	done

	sudo systemctl start fahclient.service
	sudo systemctl status --full fahclient.service
	date >> log.txt
	echo 'Ended job' >> log.txt

view raw compress_raw_video.sh hosted with ❤ by GitHub

Screencast on Linux with PulseAudio

Choose a PulseAudio source device

$ pactl list sources
<...>

To perform capture and lossless compression with x264 and Vorbis in one go, run:

	#!/bin/sh -

	# Example values
	pulseaudio_source=0
	thread_queue_size=32
	video_size=1920x1080
	framerate=30
	record_area=$DISPLAY+0,0
	naudio_channels=2
	outfile=capture.mkv

	# Command
	ffmpeg -thread_queue_size $thread_queue_size -video_size $video_size \
	-framerate $framerate -f x11grab -i $record_area \
	-thread_queue_size $thread_queue_size -f pulse -ac $naudio_channels \
	-channel_layout stereo -i $pulseaudio_source -c:v libx264 -qp 0 \
	-preset ultrafast -c:a libvorbis $outfile

view raw linux_pulseaudio_x264_lossless_vorbis.sh hosted with ❤ by GitHub

With microphone

Close pavucontrol and run below to load loopback module

$ pactl load-module module-loopback latency_msec=1

Open pavucontrol again, and configure using below pictures as guide

Playback tab

Substitute below level to 70%

Recording tab

Substitute below level Built-in Audio Analog Stereo to 70%

Input Devices tab

Then repeat steps above.

Video to GIF

Manually extract portion of video into a video file adding extra stuff like effects, subtitles, merges with images, etc.

Resize video to desired size of GIF, using only one coordinate

$ ffmpeg -i video_source.mp4 -vf scale=-1:720 video_source_resized.mp4

If you get an error about audio encoding, just copy/reencode the same audio

$ ffmpeg -i video_source.mp4 -vf scale=-1:720 -c:a copy video_source_resized.mp4

File extension mp4 refers only to container, is not that relevant. Use whatever the original video extension is.

Finally, convert to GIF

$ ffmpeg -i video_source_resized.mp4 -b 2048k video.gif

Online

https://imgur.com/vidgif. Video must be hosted elsewhere. Creates both video and GIF; to get GIF add .gif to the URL.
https://giphy.com/
http://www.video2gif.org/about.html
https://imgflip.com/

Trim/cut video to video

$ in=/path/to/input/file
$ out=/path/to/output/file

Trim/cut from minute 18:20 to 18:30. Try following commands in descending order of preference/priority.

Fast without re-encoding

$ ffmpeg -i $in -ss 00:18:20 -c:v copy -c:a copy -to 00:18:30 $out

About as fast as previous, no re-encoding

$ ffmpeg -ss 00:18:20 -i $in -c:v copy -c:a copy -t 10 $out

Slower, more accurate, with re-encoding

$ ffmpeg -ss 00:18:20 -i $in -t 10 $out

Slower, CPU-intensive, more accurate, with re-encoding

$ ffmpeg -i $in -ss 00:18:20 -to 00:18:30 $out

Short Python script

	from datetime import timedelta
	from subprocess import Popen

	'''ffmpeg_trim_cut_video2video.py

	Edit section User data, and uncomment one of the commands `cmd`
	'''

	# User data
	tstart = timedelta(minutes=18, seconds=20, milliseconds=0)
	tend = timedelta(minutes=18, seconds=30, milliseconds=0)
	src = 'video_src.mkv'
	output = 'output.mkv'
	ffmpeg = 'ffmpeg'

	tdelta = tend - tstart
	tstart = str(tstart)
	tend = str(tend)
	tdelta = str(tdelta.total_seconds())

	cmd = (ffmpeg, '-i', src, '-ss', tstart, '-c:v', 'copy', '-c:a', 'copy', '-to',
	tend, output)
	#cmd = (ffmpeg, '-ss', tstart, '-i', src, '-c:v', 'copy', '-c:a', 'copy', '-t',
	# tdelta, output)
	#cmd = (ffmpeg, '-ss', tstart, '-i', src, '-t', tdelta, output)
	#cmd = (ffmpeg, '-i', src, '-ss', tstart, '-to', tend, output)

	proc = Popen(cmd)
	print('Exit code: {}'.format(proc.wait()))

view raw ffmpeg_trim_cut_video2video.py hosted with ❤ by GitHub

Misc

Use H.264 because of its good quality, size and compatibility.
Use MP4 if using audio compression, otherwise use Matroska.
When dealing with raw audio (PCM) specify the format with the -f option (for example).
Don’t try to map audio channels without re-encoding (info).

Videos with wrong orientation in metadata

Sometimes when recording from a device like a cellphone or a tablet, the resulting video will play as if it was recorded in portrait mode

$ ffprobe VID_20160924_012917.mp4
ffprobe version 2.8.7 Copyright (c) 2007-2016 the FFmpeg developers
  built with gcc 5.3.1 (GCC) 20160406 (Red Hat 5.3.1-6)
  libavutil      54. 31.100 / 54. 31.100
  libavcodec     56. 60.100 / 56. 60.100
  libavformat    56. 40.101 / 56. 40.101
  libavdevice    56.  4.100 / 56.  4.100
  libavfilter     5. 40.101 /  5. 40.101
  libavresample   2.  1.  0 /  2.  1.  0
  libswscale      3.  1.101 /  3.  1.101
  libswresample   1.  2.101 /  1.  2.101
  libpostproc    53.  3.100 / 53.  3.100
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'VID_20160924_012917.mp4':
  Metadata:
    major_brand     : mp42
    minor_version   : 0
    compatible_brands: isommp42
    creation_time   : 2016-09-24 06:33:43
  Duration: 00:04:24.47, start: 0.000000, bitrate: 12234 kb/s
    Stream #0:0(eng): Video: h264 (Constrained Baseline) (avc1 / 0x31637661), yuv420p, 1920x1080, 12130 kb/s, SAR 1:1 DAR 16:9, 30.33 fps, 30 tbr, 90k tbn, 180k tbc (default)
    Metadata:
      rotate          : 90
      creation_time   : 2016-09-24 06:33:43
      handler_name    : VideoHandle
    Side data:
      displaymatrix: rotation of -90.00 degrees
    Stream #0:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, mono, fltp, 96 kb/s (default)
    Metadata:
      creation_time   : 2016-09-24 06:33:43
      handler_name    : SoundHandle

Below command fixes it without re-encoding.

$ ffmpeg -i VID_20160924_012917.mp4  -c copy -metadata:s:v:0 rotate=0 output.mp4

Map video and audio streams into single output

Merge the video stream of normal_export.mkv and audio stream of normal_export_Audio_1.mkv into a single container MKV normal_export_final.mkv.

$ ffmpeg -i normal_export.mkv  -i normal_export_Audio_1.mkv -map 0:0 -map 1:1 -c:v copy -c:a copy normal_export_final.mkv