Formats AV
Index
| General |
|||||||||||||||||
| Contenidors / Containers
(MUX) |
|||||||||||||||||
|
|||||||||||||||||
General
|
|
Contenidors / Containers (MUX)
| Container |
file ext | file (UNIX) | mime type |
owner | base | brand
('ftyp') |
video
codecs |
audio
codecs |
stream | usage |
|
|---|---|---|---|---|---|---|---|---|---|---|---|
| file |
transmission |
||||||||||
| 3GP | video/3gpp |
3GPP | ISO Base media file format (isobmff) (MPEG-4 part 12)(MP4reg) | 3g?? |
|
|
x | ||||
| MOV | .mov | ISO Media, Apple QuickTime movie | video/quicktime | Apple | qt |
x | x | x | |||
| MJ2 (Motion JPEG2000) | |||||||||||
| MP4 | .mp4 | ISO Media, MPEG v4 system | MPEG |
|
|
x | |||||
| DVB-FF |
.dvb |
video/vnd.dvb.file | DVB |
|
|||||||
| ASF | Microsoft | x | x | x | |||||||
| AVI | .avi | Microsoft | x | x | - | ||||||
| DivX | DivXNetworks | x | x | - | |||||||
| Flash Video | .flv | Macromedia Flash Video | video/x-flv | Adobe |
|
|
RTMP | ||||
| .f4v |
video/mp4 | ISO base media file |
|
|
|||||||
| Matroska
(wp) |
.mkv .mka .mks |
CoreCodec | MCF |
|
|
x | |||||
| MCF | BSD | x | x | x | |||||||
| MPEG PS | .mpg .mpeg .vob |
MPEG sequence, v2, program multiplex | MPEG |
|
|
? | |||||
| MPEG TS | .ts .trp |
MPEG transport stream data (TeX font metric data) |
MPEG |
|
x | ||||||
| BDAV MPEG
TS |
.m2ts |
MPEG |
|
||||||||
| MXF | .mxf | SMPTE | |||||||||
| OGG (RFC 3533) |
Xiph | |
|
x | |||||||
| OGM (RFC 3534) |
|
|
|||||||||
| RMVB | Real Networks | RealVideo | x | x | |||||||
| SVI (AVI) | .svi | Samsung | AVI | XVID 208x176 15fps HAS_INDEX IS_INTERLACED TRUST_CKTYPE |
mpga 2ch 44100Hz 128Kbps |
||||||
| VOB (MPEG PS) |
MPEG |
|
|
||||||||
| WEBM |
.webm |
video/webm audio/webm |
Google |
Matroska |
VP8 |
Vorbis |
|||||
| Container |
file ext | file (UNIX) | mime type |
owner | base | video
codecs |
audio
codecs |
stream | |||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Conversió / Conversion
- Programari/software
- HOWTO Convert video files
- MTS
files (AVCHD: H.264 + AAC on
MPEG2-TS, from camera)
- Concatenació /
Concatenation
- cat (second step is necessary to avoid
discontinuities: "Continuity check failed for pid ...
expected ... got ...")
- cat 00650.MTS 00651.MTS 00652.MTS > temporal.mts
- avconv -i temporal.mts -c:v copy -c:a copy -f mpegts tots.mts
- avconv -i temporal.mts -c:v copy -c:a copy tots.mp4
- find
all files between two dates (2013-08-9 to 2013-08-11)
- cat $(find . -newermt 2013-08-9 ! -newermt 2013-08-12 -printf "%T@ %p\n" | sort -n | awk '{print $2}' | paste -s -d ' ') > 2013-08-09_11.mts
- avconv -i 2013-08-09_11.mts -c:a copy -c:v copy 2013-08-09_11.mp4
- melt
- libav: Concatenació / Concatenation
- ffmpeg: Concatenació / Concatenation
- aplicacions d'edició / gui edition applications
- Howtos
- Thread: tsMuxer and joining
m2ts files
- MPLS playlist
- Thread: How to merge multiple
.mts files (canon HF) into one single .mts file
- tsmuxer
- Thread: tsMuxer and joining
m2ts files
- cat (second step is necessary to avoid
discontinuities: "Continuity check failed for pid ...
expected ... got ...")
- DLNA
- Concatenació /
Concatenation
- Subtítols / Subtitles
- Subtitle manipulation tools for Linux
- [vdr]
How
to extract HD dvb subtitles (and possibly convert to srt)?
- vdrsubrip
uses:
- dependencies
- ProjectX to extract subtitles (urpmi projectx)
- BDSup2Sub
(doom9)
(digital-digest)
to convert to images
- svn --username anonymous co http://javaforge.com/svn/bdsup2sub/BDSup2Sub/ BDSup2Sub
- ImageMagick to improve images
- gocr for OCR
- usage:
- edit vdrsubrip:
- PROJX='/usr/share/java/ProjectX.jar'
- BDSUP2SUB='~/src/BDSup2Sub/BDSup2Sub.jar'
- edit vdrsubrip:
- dependencies
- vdrsubrip
uses:
- [linux-dvb]
how
to extract subtitles in text format?
- son2srt
- Extraction with transcode
- Extraction with ffmpeg
- Eines / Tools:
- rippers:
- KSubtitleRipper (gui)
- subtitleripper
(uses: transcode, gocr)
- compilation
- urpmi libnetpbm-devel libpng-devel
- compilation
- rippers:
- wget ...
- vobsub.c:
- rename getline to getlinia (or anything else)
- Makefile:
- INCLUDES := -I /usr/include/netpbm
- #LIBS += -lppm
LIBS += -lnetpbm
- make
- subtitle2pgm
- Gnome Subtitles
- KSubtile
- SubtitleEditor
- Gaupol
- GSubEdit
- Jubler
- sub2srt
- Subtitles
- compensate the display time in 3 seconds:
- subs -i -b 3 subtitle_file.sub
- 24fps -> 25 fps:
- subs -i -a 24/25 subtitle_file.sub
- time from file1 and text from file2:
- subs -z subtitle_file1.sub subtitle_file2.sub
- compensate the display time in 3 seconds:
- MPEG4IP
- Using MJPEG-tools
- TSDuck
- TSDuck is also a large library to manipulate MPEG transport streams. C++, Java and Python developers can use bindings in their languages.
- Using mencoder (mplayer)
- Using transcode (tccat,
tcextract):
- urpmi
- Subtitle
manipulation tools for Linux
- DVD -> /home/user/rip/vob
- tccat -i /home/user/rip/vob | tcextract -x ps1 -t vob -a 0x22 > subs-en
- other languages:
- to determine the sid for the preferred subtitles:
- mplayer -dvd-device /space/st-tng/disc1/ -dvd 1 -vo null -ao null -frames 0 -v 2>&1 | grep sid
- if you want language with sid 7, then add it to
0x20:
- tccat -i /home/user/rip/vob | tcextract -x ps1 -t vob -a 0x27 > subs-it
- to determine the sid for the preferred subtitles:
- convert them to images:
- subtitle2pgm -o english -c 255,0,255,255 < subs-en
- OCR:
- pgm2txt english
- create srt file:
- srttool -s -w < english.srtx > english.srt
- Estabilització / Stabilization
- libav
(fork from ffmpeg)
- Història / History
- Instal·lació / Installation
- Compilació
/ Compilation
- pipe
-
syntax
meaning
example
pipe:0
stdin
... | avconv -i pipe:0 ...
pipe:1
stdout
avconv ... pipe:1 | tee -a fitxer1 fitxer2
pipe:2
stderr
- avconv -f video4linux2 -i /dev/video0 -s 320x240 -vcodec libx264 -f mpegts pipe:1 | avplay -i pipe:0
- avconv -i tvc_794_20140821_1709.ts -c copy -f mpegts - | avplay -i -
- mkfifo
- mkfifo toto.ts
- avconv -i ... -f mpegts - > toto.ts
- [avconv -i ... -f mpegts -y toto.ts]
- avplay toto.ts
- simultaneous
2 outputs to 2 fifos (does not work)
- mkfifo v1.ts v2.ts
- ffmpeg -i sintel-1280-surround.mp4 -c:v libx264 -b:v 500k -s 640x360 -aspect 16:9 -an -f mpegts -y v1.ts -c:v libx264 -b:v 1500k -s 1280x720 -aspect 16:9 -f mpegts -y v2.ts
- other examples
-
- General
- Filtres / Filters
- ffmpeg filters
- Info about input (only available when specifying an
output)
- avconv -i input -filter:v showinfo output
- Tall / Cut (5s)
- avconv -threads auto -i toto.MTS -t 0:00:05.0 -c:v copy -c:a copy -f mpegts toto_5s.MTS
- Concatenació /
Concatenation
- see also
- How can I join video files? (mkfifo)
- MTS files:
- avplay -i "concat:00185.MTS|00186.MTS|00187.MTS"
- MPEG-DASH
segments:
- avplay -i "concat:segment_init.mp4|segment_1.m4s|segment_2.m4s"
- Entrellaçat / Interlaced
- Interlaced video looks very bad when encoded with avconv, what is wrong?
- -flags +ilme+ildct
- -flags +alt?
- -top 0/1
- deinterlace
- -filter:v yadif
- Rotació i rescalatge /
Rotation and rescale
- avconv -threads auto -i toto.mts -filter:v transpose=clock toto_clock.mp4
- resize and then rotation:
- resize, rotation and pad:
- avconv -threads auto -i 00189_5s.MTS -filter:v "scale=w=576:h=-1,transpose=clock,pad=width=720:height=576:x=(ow-iw)/2:y=0:color=black" -f mpegts 00189_5s_clock.MTS
- avconv -threads auto -i 00189_5s.MTS -filter:v "scale=w=576:h=-1,transpose=clock,pad=width=720:height=576:x=(ow-iw)/2:y=0:color=black" -strict experimental 00189_5s_clock.mp4
- Relació
d'aspecte / Aspect ratio (ffmpeg)
- Sobreimpressió / Overlay
- Transport
Stream (TS) (see also ffmpeg
TS)
- mpegts
- TS streaming
- change PAT/PMT
rate (default: every 100ms):
- libavformat/mpegtsenc.c
- PAT_RETRANS_TIME 100
- libavformat/mpegtsenc.c
- MP4 (see also ffmpeg MP4)
- Streaming (see also ffmpeg
streaming)
-
-re option is needed when input is a file
server client (192.168.1.100)
output option: -f ...
(check that the port is not closed by the firewall) RTP / SDP rtp
unicast
- (?) avconv -re -i toto.mp4 -c:a copy -c:v copy -f rtp rtp://192.168.1.100:1234
multicast
- (?) avconv -re -i toto.mp4 -c:a copy -c:v copy -f rtp rtp://234.1.2.3:1234
TS over UDP
mpegts
unicast - avconv -re -i toto.mp4 -bsf h264_mp4toannexb -c:a copy -c:v copy -f mpegts "udp://192.168.1.100:1234?ttl=10&buffer_size=65536&pkt_size=1316"
- avconv -re -i toto.mp4 -f mpegts "udp://192.168.1.100:1234?ttl=10&buffer_size=65536&pkt_size=1316"
multicast - avconv -re -i toto.mp4 -c:v copy -c:a copy -bsf h264_mp4toannexb -f mpegts "udp://234.1.2.3:1234?ttl=10&buffer_size=65536&pkt_size=1316"
- avconv -re -i toto.mp4 -f mpegts "udp://234.1.2.3:1234?ttl=10&buffer_size=65536&pkt_size=1316"
- server
- RTP / SDP
- from file
- avconv -re -i ...
- from camera
- avconv -f video4linux2 -i /dev/video0 -s 320x240 -vcodec libx264 -f rtp rtp://192.168.1.100:1234
- send generated sdp (shown on screen when
launching previous command) to client as
toto.sdp
- v=0
o=- 0 0 IN IP4 127.0.0.1
s=No Name
c=IN IP4 192.168.1.100
t=0 0
a=tool:libavformat 55.0.1
m=video 1234 RTP/AVP 96
a=rtpmap:96 H264/90000
a=fmtp:96 packetization-mode=1
- v=0
- from file
- TS over UDP
(use pkt_size=1316
to allocate 7 TS packets into one IP packet:
188*7=1316 < 1500):
- (see also OpenCaster tsudpsend)
- unicast (-re option needed if input is a file)
- avconv -i ... -f mpegts -muxrate 1200000 udp://unicast_destination_address:1234?buffer_size=65536&pkt_size=1316
- no transcoding (H.264/AAC) (depending on
original mp4 file, the option -bsf may be
necessary)
- avconv -re -i toto.mp4 -bsf h264_mp4toannexb -c:a copy -c:v copy -f mpegts udp://192.168.1.100:1234?ttl=10&buffer_size=65536&pkt_size=1316
- transcode from mp4 (h.264/aac) to mpeg2ts
(mpeg2video/mp2):
- avconv -re -i toto.mp4 -f mpegts udp://192.168.1.100:1234?ttl=10&buffer_size=65536&pkt_size=1316
- multicast
- avconv -i ... -f mpegts -muxrate 1200000 udp://multicast_destination_address:1234?ttl=10&buffer_size=65536&pkt_size=1316
- implicit transcoding to mpeg2video/mp2:
- avconv -re -i toto.mp4 -f mpegts udp://234.1.2.3:1234?ttl=10&buffer_size=65536&pkt_size=1316
- no transcoding:
- avconv -re -i toto.mp4 -c:v copy -c:a copy -bsf h264_mp4toannexb -f mpegts udp://234.1.2.3:1234?ttl=10&buffer_size=65536&pkt_size=1316
- RTP / SDP
- client (192.168.1.100)
- RTP / SDP
- unicast
- just play:
- avplay toto.sdp
- write the stream to a TS file with TS
constant bit-rate (e.g. 80kbps):
- avconv -i toto.sdp -c:v copy -y -f mpegts -muxrate 80000 toto.ts
- just play:
- multicast
- ...
- unicast
- TS over UDP (see also ffmpeg
UDP):
- -infbuf option is needed to avoid errors (see also fifo_size for ffmpeg)
- unicast
- avplay -infbuf udp://@:1234
- avconv ... -i udp://...fifo_size=...
- multicast
- avplay -infbuf -i udp://@234.1.2.3:1234
- RTP / SDP
-
- H.264
/ libx264
(see also ffmpeg with x264)
- libx264
- GOP size: 100
- -g 100
- GOP size: 100
- baixa latència / low latency
- x264 presets:
- avconv -preset ultrafast
- avconv presets
- /usr/local/share/avconv/libx264-*.avpreset
- avconv -pre:v 1080p50_60
- /usr/local/share/avconv/libx264-toto_preset.avpreset
(com es fa amb un fitxer local? quines opcions hi ha
disponibles?)
- slices=2
x264-params=cabac=0
...
- avconv ... -pre:v toto_preset ...
- slices=2
- /usr/local/share/avconv/libx264-*.avpreset
- baseline profile del vídeo redimensionat a 360x* ("-1"
fa que es mantingui la relació), a 800 kbps:
- avconv -i toto.ts -filter:v scale=360:-1 -b:v 800k -profile:v baseline toto.mp4
- Problemes / Problems
- "[libx264] lookaheadless mb-tree requires intra refresh or infinite keyint"
- libx264
- Using ffmpeg
- Info
- ffmpeg-libav-tutorial (Leandro Moreira)
- FFmpeg_Book (pdf) (Astro Electronic)
- Bird trails
- Docs
- Documentation
- All documentation
- Filter
documentation
- Audio
sources
ffplay -f lavfi -i "sine=frequency=5000:sample_rate=44100, aformat=channel_layouts=stereo"
- Video
sources
- mandelbrot, allrgb, allyuv, color, colorspectrum, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
ffplay -f lavfi -i mandelbrotffmpeg -f lavfi -i testsrc=size=864x480- ffmpeg -lavfi 'testsrc2;sine' toto.mov
ffmpeg -f lavfi -i "testsrc=duration=10:size=640x360, format=pix_fmts=yuv420p" -f lavfi -i anoisesrc=duration=10:colour=pink:amplitude=0.5 -ac 2 toto.mp4- avsynctest
- openclsrc
- ...
- Audio
sources
- wiki
- Acceleració per
maquinari / Hardware acceleration
- HWAccelIntro
- Usage
- Detection
- check whether ffmpeg has been compiled
with nvenc support:
ffmpeg -codecs | grep nvenc
- check whether ffmpeg has been compiled
with nvenc support and system allows
usage of nvenc:
ffmpeg -f lavfi -i nullsrc -c:v h264_nvenc -t 00:00:00 -f null /dev/null- all ok (exit with 0)
Output file is empty, nothing was encoded (check -ss / -t / -frames parameters if used)
- ffmpeg has not been compiled
with cuda (exit with 1)
- Unknown encoder 'h264_nvenc'
- system is not able to deal
with cuda (exit with 1)
- [h264_nvenc @ 0x762c00] dl_fn->cuda_dl->cuInit(0) failed -> CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected
- all ok (exit with 0)
- echo $?
- example script:
#!/bin/bash
function is_ffmpeg_nvenc_available {
# check whether ffmpeg has h264_nvenc coder and system is able to use it
ffmpeg -f lavfi -i nullsrc -c:v h264_nvenc -t 00:00:00 -f null /dev/null 1>/dev/null 2>&1
result=$?
if ! (( result ))
then
# ffmpeg works ok
echo 1
else
# ffmpeg exits with 1 (error)
echo 0
fi
}
nvenc_available=$(is_ffmpeg_nvenc_available)
echo "nvenc_available: ${nvenc_available}"
- check whether ffmpeg has been compiled
with nvenc support:
- Encoding
ffmpeg -y -f lavfi -i "testsrc=duration=20:size=1920x1080, format=pix_fmts=yuv420p" -f lavfi -i anoisesrc=duration=10:colour=pink:amplitude=0.5 -ac 2 -c:v h264_nvenc toto_nvenc.mp4- ffmpeg -f lavfi -i nullsrc -c:v h264_nvenc -gpu list -f null -
- Decoding
ffmpeg -hwaccel cuda -i toto.mp4
- Filtering
- Detection
- Output from ./configure
system info
vdpauinfo
Mageia dependencies
nv-codec-headers ? libvdpau-devel libva-devel lib64omxil-bellagio-devel
CentOS/Alma/Rocky dependencies
nv-codec-headers
Debian/Raspbian dependencies
libomxil-bellagio-dev provided pkg-config
ffnvcodec
auto-detect (no need to specify --enable-...)
--enable-vdpau --enable-vaapi --enable-omx --enable-omx-rpi External libraries providing
hardware acceleration:
- ffmpeg -hwaccels
cuda cuvid ffnvcodec nvdec nvenc v4l2_m2m vdpau vaapi omx
Enabled decoders AV1
H.263
h263_v4l2m2m
H.264
h264_cuvid
h264_v4l2m2m
HEVC
hevc_cuvid
hevc_v4l2m2m
MJPEG
mjpeg_cuvid
MPEG-1
mpeg1_cuvid
mpeg1_v4l2m2m
MPEG-2
mpeg2_cuvid
mpeg2_v4l2m2m
MPEG-4
mpeg4_cuvid
mpeg4_v4l2m2m
VC1
vc1_cuvid
vc1_v4l2m2m
VP8
vp8_cuvid
vp8_v4l2m2m
VP9
vp9_cuvid
vp9_v4l2m2m
Enabled hwaccels
- ./configure --list-hwaccels
AV1
av1_nvdec
av1_vaapi
H.263
h263_vaapi
H.264
h264_nvdec
h264_vdpau h264_vaapi
HEVC
hevc_nvdec
hevc_vdpau hevc_vaapi
MJPEG
mjpeg_nvdec
mjpeg_vaapi
MPEG-1
mpeg1_nvdec
mpeg1_vdpau
MPEG-2
mpeg2_nvdec
mpeg2_vdpau mpeg2_vaapi
MPEG-4
mpeg4_nvdec
mpeg4_vdpau mpeg4_vaapi
VC1
vc1_nvdec
vc1_vdpau vc1_vaapi
VP8
vp8_nvdec
vp8_vaapi
VP9
vp9_nvdec
vp9_vdpau vp9_vaapi
WMV3
wmv3_nvdec
wmv3_vdpau wmv3_vaapi
Enabled encoders H.263
h263_v4l2m2m
H.264
h264_nvenc, nvenc_h264 h264_v4l2m2m
h264_vaapi h264_omx
HEVC
hevc_nvenc, nvenc_hevc hevc_v4l2m2m
hevc_vaapi
MJPEG
mjpeg_vaapi
MPEG-1
MPEG-2
mpeg2_vaapi
MPEG-4
mpeg4_v4l2m2m
mpeg4_omx
VP8
vp8_v4l2m2m
-
configure option
(./configure --list-hwaccels)
codecs
libav
ffmpeg
mjpeg mpeg1 mpeg2
mpeg4
h263
h264
hevc vc1
vp8 vp9 wmv3
--disable-dxva2
x
CUDA
x x x x
x x x x x x OpenMAX --enable-omx
--enable-omx-rpi
v4l2_m2m
d d ed ed ed
d
VA API
--enable-vaapi
[--enable-vaapi]
--enable-hwaccel=h264_vaapix
x
x
x
x
x x
x x x
VDA
--enable-vda
--enable-hwaccel=h264_vda
x
VDPAU
--enable-hwaccel=h264_vdpau
x x
x
x
x x
x
- Creating
multiple
outputs
... -f tee -flags +global_header -map 0:v -map 0:a "[...]...|[...]..."
- Problemes /
Problems
- NOTE: if -flags +global_header is not specified and one of the outputs is RTMP, it will appear without video profile in /stats (nginx-rtmp-module) and HTML5 player with video.js will not be able to start it
[tee @ 0x80cf20] Non-monotonous DTS in output stream 0:0; previous: 566, current: 561; changing to 567. This may result in incorrect timestamps in the output file.
- Examples:
- 1 transcoding and 2 output muxers: HLS + RTMP
- ffmpeg -i ... \
-c:a ... -c:v ... \
-f tee -flags +global_header -map 0:v -map 0:a \
"[hls_time=2:hls_list_size=0:hls_wrap=0:start_number=1]toto.m3u8|[f=flv]rtmp://localhost:1935/..."
- ffmpeg -i ... \
- 2 transcodings: one with HLS+RTMP, the other one
with HLS+SAP
- ffmpeg -i ... \
-vsync vfr -c:v h264 -preset medium -b:v 1000000 -vf scale=w=720:h=576 -g 25 -profile:v main -level 30 -keyint_min 25 -sc_threshold 0 \
-c:a aac -strict -2 -b:a 128000 -ac 2 \
-f tee -flags +global_header -map 0:v -map 0:a \
"[hls_time=2:hls_list_size=0:hls_wrap=0:start_number=1]sd.m3u8|[f=flv]rtmp://localhost:1935/sd/..." \
-vsync vfr -c:v h264 -preset medium -b:v 2000000 -vf scale=w=1280:h=720 -g 25 -profile:v high -level 31 -keyint_min 25 -sc_threshold 0 \
-c:a aac -strict -2 -b:a 128000 -ac 2 \
-f tee -flags +global_header -map 0:v -map 0:a \
"[hls_time=2:hls_list_size=0:hls_wrap=0:start_number=1]hd.m3u8|[f=sap]sap://234.1.2.3" \
- ffmpeg -i ... \
- 1 transcoding and 2 output muxers: HLS + RTMP
- How to speed up / slow down a video time lapse, etc.
- Option '-sameq' does NOT mean 'same quality'
- How to take multiple screenshots to an image (tile, mosaic)
- Create a mosaic out of several input videos
- Filtering
guide
- Acceleració per
maquinari / Hardware acceleration
- options
- more options
- ffmpegX options
- ffplay -loglevel
verbose ...
- ffplay -v {quiet,panic,fatal,error,warning,info,verbose,debug} [-vf showinfo] ...
- Protocol 'rtp' not on whitelist 'file,crypto'!
- ffplay -safe 0 -protocol_whitelist rtp,udp,file,http,https,tcp,tls -i toto.sdp
- keyboard
(cursor over window showing video)
- q ESC quit
f toggle full screen
p,SPC pause
a cycle audio channel in the current program
v cycle video channel
t cycle subtitle channel in the current program
c cycle program
w cycle video filters or show modes
s activate frame-step mode
left/right seek backward/forward 10 seconds
down/up seek backward/forward 1 minute
pg down/pg up seek backward/forward 10 minutes
mouse right click seek to percentage in file corresponding to fraction of width
- q ESC quit
- webcam
- Capture / Webcam
- Dispositius / Devices
v4l2-ctl --list-devicesv4l2-ctl -d /dev/video1 -Lv4l2-ctl -c <option>=<value>
- Visualització / View
- ffmpeg [-v ...] [options_for_input_file_0 [-i input_file_0]] [options_for_input_file_1 [-i input_file_1]] [-map 0:x] [-map 1:y] [options_for_output_file_0 [output_file_0]] [options_for_output_file_1 [output_file_1]] [options -newaudio] [options -newvideo]
- keyboard actions
(type "?" to see help)
- ? show this help
+ increase verbosity
- decrease verbosity
c Send command to first matching filter supporting it
C Send/Que command to all matching filters
D cycle through available debug modes
h dump packets/hex press to cycle through the 3 states
q quit
s Show QP histogram
- ? show this help
- Generic
options (show available ...):
- ffmpeg -formats
- ffmpeg -codecs
- ffmpeg -filters
- ffmpeg -pix_fmts (show available pixel formats: -pixel_format)
- ffmpeg -sample_fmts
- general parameters
- -d daemonize
(must be the very first option)
- -stream_loop
<number> (0: no loop; -1: infinite loop; n: number
of loops)
- -i input file
- Video
sources (synthetic videos)
- [ffplay,ffmpeg] -f lavfi -i testsrc=size=1280x720 ...
- Video
sources (synthetic videos)
- -f container/ES_type
- -loglevel, -v {quiet,panic,fatal,error,warning,info,verbose,debug,trace}
- -d daemonize
(must be the very first option)
- flags / fflags
- complete list of ffmpeg flags / commands
- -flags
- global_header
- loop
- ...
- -fflags
- nobuffer
- genpts
- igndts
- ...
- video parameters
- -c:v
video_codec
- -b:v video_bit-rate
- -b video bit-rate
- -bt:v video_bit-rate tolerance
- -r frame-rate
- -s size (see also scale filter)
- -aspect aspect_ratio
- -c:v
video_codec
- audio parameters
- -c:a
audio_codec
- -b:a audio_bit-rate
- -ac number_of_audio_channels
- -ar audio_sampling_rate
- -c:a
audio_codec
- -map
input_file:index (when using map option, all
output streams to be mapped must be specified; if output
file can contain both audio and video, e.g. mp4, two -map
must be specified)
- examples (*):
- ... -map 0:0 -map 0:2 ... toto.mp4
- stream #0:0 (video) to toto.mp4 video track
- stream #0:2 (2nd audio) to toto.mp4 main audio track
- .. -map 0:0 -map 0:2 -map 0:1 ... toto.mp4
-newaudio
- stream #0:0 (video) to toto.mp4 video track
- stream #0:2 (2nd audio) to toto.mp4 main audio track
- stream #0:1 (1st audio) to toto.mp4 secondary audio track
- ... -map 0:3 -map 2:4 out_a.mp2 out_b.mp3
- stream #0.3 to out_a.mp2
- stream #2.4 to out_b.mp3
- ffmpeg -i first_input.mp4 -i second_input.mp4 -i
third_input.mp2 -map 0:0 -map 0:2 -map 1:1 -map 1:1
-map 2:1 toto.mp4 -newaudio out_a.mp2 out_b.mp3
input
output
first_input.mp4 contains:
- 0:0: video track
- 0:1: audio track
- 0:2: audio track
-i first_input.mp4 second_input.mp4 contains:
- 1:0: video track
- 1:1: audio track
-i second_input.mp4 third_input.mp2 contains:
- 2:0: audio track
- 2:1: audio track
-i third_input.mp2 -map 0:0 toto.mp4 toto.mp4 will contain the following tracks:
- 0:0: video from first_input
- 0:1: audio from second audio track from first_input
- 0:2: audio from first audio track from second_input
-map 0:2 -map 1:1 -newaudio -map 1:1 out_a.mp2 out_a.mp2 will contain the following tracks:
- 1:0: audio from first audio track from second_input
-map 2:1 out_b.mp3 out_b.mp3 will contain the following tracks:
- 2:0: audio from second audio track from third input
- TS program selection
- ... -map 0:0 -map 0:2 ... toto.mp4
- examples (*):
- pipe
- multiple inputs
- Switch (mixer)
- múltiples sortides / multiple outputs
- sample formats
- list of available formats
- ffmpeg -sample_fmts
- libswresample (libav: libavresample)
- conversion from s16p to s16:
- How to convert sample
rate from AV_SAMPLE_FMT_FLTP to AV_SAMPLE_FMT_S16?
- libswresample
- [Libav-user] Audio conversion from planar to non-planar formats
- (?) fffmpeg ... -sample_fmt s16 ...
- How to convert sample
rate from AV_SAMPLE_FMT_FLTP to AV_SAMPLE_FMT_S16?
- list of available formats
- Start from time (-ss) hh:mm:ss[:ff],
during duration (-t):
- ffmpeg ... -ss hh:mm:ss[:ff] -t hh:mm:ss[.dd] ...
- specify -ss before -i to avoid anything but key frames: much faster (*)
- "-ss seconds" for an entire number of seconds
- filtres / filtering
- FFmpeg Filters Documentation
- Filtering guide (wiki)
- chained filters
- simple
- Filtergraph,Chain,Filter relationship
- -vf "scale=320:-1, drawtext=fontfile=/usr/share/fonts/google-droid/DroidSans.ttf: timecode='00\:00\:00\:00': r=24: x=(w-tw)/2: y=h-(2*lh): fontsize=50: fontcolor=white"
- complex
- -filter_complex
- simple
- list of available filters:
- ffmpeg -filters
- audio
- video
- ffmpeg -filter:v ...
- ffmpeg -vf ...
- show information (see also ashowinfo for audio) (n pts pts_time pos fmt sar s i iskey type checksum plane_checksum mean stdev)
- remove the logo
- ffplay -i tv3.m2v -vf 'delogo=x=620:y=40:w=50:h=50:band=10:show=0'
- add a logo (see OpenCaster):
- ffmpeg -vf ...overlay...
- ffmpeg -i ${video_src} -i ${logo_src} \
-filter_complex "[0:v]scale=320:-1[output]; [1:v][output]scale2ref=iw/8:-1[logo][output]; [output][logo]overlay=main_w/10:main_h/10[output]" -map "[output]" -map 0:a -c:v h264 -c:a aac -f tee "[f=mp4]output1_tl.mp4|[f=mpegts]output1_tl.ts" \
-filter_complex "[0:v]scale=640:-1[output]" -map "[output]" -map 0:a -c:v h264 -c:a aac -f tee "[f=mp4]output2.mp4"
- display subtitles
- dep
- Mageia
- urpmi libass-devel
- Mageia
- ./configure ... --enable-libass
- ffmpeg -vf 'subtitles'
- dep
- scale (resize)
- PIP
- ffplay -i 324.m2v -vf 'movie=tv3.m2v, scale=180:-1 [petit]; [in][petit] overlay=16:16 [out]'
- ffplay -i 794.ts -vst 11 -ast 12 -vf "movie=794.ts:s=0, scale=180:-1 [v1]; [in][v1] overlay=16:16 [out]"
- ffmpeg -i 794.ts -filter_complex "[0:v:0]scale=width=240:height=180[petit]; [0:v:1][petit] overlay" -f mpegts - | ffplay -
- ffmpeg -i 794.ts -b:v 2M -filter_complex "[0:v:0]scale=width=240:height=180[petit]; [0:p:803:v][petit] overlay" -f mpegts - | ffplay -
- mosaic
- ffmpeg -i 794.ts -b:v 4M -filter_complex "nullsrc=size=1280x720 [base]; [0:v:0] scale=width=640:height=360 [upperleft]; [0:v:1] scale=width=640:height=360 [upperright]; [0:v:2] scale=width=640:height=360 [lowerleft] ; [0:v:3] scale=width=640:height=360 [lowerright]; [base][upperleft] overlay [tmp1]; [tmp1][upperright] overlay=shortest=1:x=640 [tmp2]; [tmp2][lowerleft] overlay=shortest=1:y=360 [tmp3]; [tmp3][lowerright] overlay=shortest=1:x=640:y=360" -f mpegts - | ffplay -
- drawtext
- ffmpeg has to be compiled with
--enable-libfreetype, and, optionally,--enable-fontconfig(to allow not specifying font) - Burnt
in
timecode
- Dependencies
- Mageia
- urpmi google-droid-sans-fonts
- fontfile=/usr/share/fonts/google-droid/DroidSansMono.ttf
- urpmi fonts-ttf-droid
- urpmi google-droid-sans-fonts
- CentOS
- sudo yum install
google-droid-sans-mono-fonts
google-droid-sans-fonts
- fontfile=/usr/share/fonts/google-droid/DroidSansMono.ttf
- sudo yum install
google-droid-sans-mono-fonts
google-droid-sans-fonts
- Ubuntu
- sudo apt-get install
fonts-droid-fallback
- fontfile=/usr/share/fonts-droid/truetype/droid/DroidSansFallback.ttf
- sudo apt-get install
fonts-droid-fallback
- Mageia
- ffplay -f lavfi -i testsrc=size=1280x720 -vf "drawtext=fontfile=${fontfile}: timecode='09\:57\:00\:00': r=25: x=(w-tw)/2: y=h-(2*lh): fontcolor=white: box=1: boxcolor=0x00000000@1"
- ffplay -f lavfi -i testsrc=size=1280x720 -vf "drawtext=fontfile=/usr/share/fonts/TTF/droid/DroidSans.ttf: timecode='09\:57\:00\:00': r=25: x=(w-tw)/2: y=h-(2*lh): fontcolor=white: box=1: boxcolor=0x00000000@1"
- ffmpeg -i sintel-1024-stereo.mp4 -vf "drawtext=fontfile=${fontfile}: timecode='00\:00\:00\:00': r=24: x=(w-tw)/2: y=h-(2*lh): fontsize=50: fontcolor=white: box=1: boxcolor=0x00000000@1" -c:a copy -c:v libx264 sintel-1024-stereo_timecode.mp4
- present time:
- ... -vf "drawtext=fontfile=${fontfile}: text='%{localtime\:%T}': r=24: x=(w-tw)/2: y=h-(2*lh): fontsize=50: fontcolor=white: box=1: boxcolor=0x00000000@1" ...
- timecode and present time:
- ffplay -f lavfi -i testsrc=size=1280x720:r=25 -vf "format=pix_fmts=yuv420p, drawtext=fontfile=/usr/share/fonts/google-droid/DroidSans.ttf: text='%{gmtime\:%T %Z}': r=25: x=(w-tw)/3*2: y=h-(2*lh): fontcolor=white: box=1: boxcolor=0x00000000@1: fontsize=50, drawtext=fontfile=/usr/share/fonts/google-droid/DroidSans.ttf: timecode='00\:00\:00\:00': r=25: x=(w-tw)/3: y=h-(2*lh): fontsize=50: fontcolor=white: box=1: boxcolor=0x00000000@1"
- ffmpeg -f lavfi -i testsrc=size=1280x720:r=25 -vf "format=pix_fmts=yuv420p, drawtext=fontfile=/usr/share/fonts/google-droid/DroidSans.ttf: text='%{gmtime\:%T %Z}': r=25: x=(w-tw)/3*2: y=h-(2*lh): fontcolor=white: box=1: boxcolor=0x00000000@1: fontsize=50, drawtext=fontfile=/usr/share/fonts/google-droid/DroidSans.ttf: timecode='00\:00\:00\:00': r=25: x=(w-tw)/3: y=h-(2*lh): fontsize=50: fontcolor=white: box=1: boxcolor=0x00000000@1" -c:v libx264 -g 12 -keyint_min 12 -sc_threshold 0 -f flv rtmp://...
- timecode and present time, with audio, to
rtmp:
-
- resize to 1280x720, downmix to stereo, add
timecode:
- ffmpeg -i big_buck_bunny_1080p_h264.mov -vf "scale=1280:-1, drawtext=fontfile=/usr/share/fonts/google-droid/DroidSans.ttf: timecode='00\:00\:00\:00': r=24: x=(w-tw)/2: y=h-(2*lh): fontsize=50: fontcolor=white: box=1: boxcolor=0x00000000@1" -ac 2 -c:a aac -strict -2 -b:a 64k -c:v libx264 -b:v 700k bbb_timecode_1280x720_700k_stereo.mp4
- Dependencies
- Dynamic text
- ffmpeg has to be compiled with
- show DVB subtitles:
- from ts file
- ffplay -i tvc_794_20140821_1709.ts -vst 2 -ast 7 -sst 23
- from live dvb (no funciona, perquè potser tzap
no conserva els subtítols)
- tzap -pr "TV3(TVC)"
- ffplay -f mpegts -i /dev/dvb/adapter0/dvr0 -vf 'subtitles=...
- from ts file
- burn DVB subtitles:
- ffmpeg -i tvc_794_20140821_1709.ts
- Program 801
Metadata:
service_name : ?TV3
service_provider: ?TVC
Stream #0:2[0x6f]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p(tv), 720x576 [SAR 64:45 DAR 16:9], max. 15000 kb/s, 25 fps, 25 tbr, 90k tbn, 50 tbc
Stream #0:7[0x70](cat): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, stereo, s16p, 158 kb/s
Stream #0:6[0x72](v.o): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, mono, s16p, 94 kb/s
Stream #0:8[0x73](ac3): Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, stereo, fltp, 192 kb/s
Stream #0:22[0x74](a.d): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, mono, s16p, 94 kb/s
Stream #0:23[0x321](cat): Subtitle: dvb_subtitle ([6][0][0][0] / 0x0006)
Stream #0:24[0x32b](eng): Subtitle: dvb_subtitle ([6][0][0][0] / 0x0006)
...
- Program 801
- ffmpeg -i tvc_794_20140821_1709.ts -filter_complex '[#0x6f][#0x321] overlay' -sn -map '#0x70' -f mpegts - | ffplay -
- ffmpeg -i 794.ts -filter_complex '[0:v][0:s:0]overlay[v]' -map [v] -map 0:a ...
- ffmpeg -i tvc_794_20140821_1709.ts
- ffmpeg -f lavfi -i amovie=toto.ts,silencedetect=noise=0.0001 -f null -
- blackframe
- ffmpeg -f lavfi -i movie=tvc_hd_20121116-2345.ts,blackframe -f null -
- segmentació / segmentation
- concatenació /
concatenation
- libav: Concatenació / Concatenation
- MTS: Concatenació / Concatenation
- Concatenate
(How
to
concatenate (join, merge) media files) (wiki)
- Concat
filter
- Does not accept -c copy
- Manually
- directly concatenable formats
- cat imput1.ts input2.ts > output.ts
- not directly concatenable formats, but
transmuxable to a concatenable format (e.g.
H.264/AAC in MP4, to TS)
- mkfifo temp1 temp2
ffmpeg -i input1.mp4 -c copy -bsf:v h264_mp4toannexb -f mpegts temp1 2> /dev/null & \
ffmpeg -i input2.mp4 -c copy -bsf:v h264_mp4toannexb -f mpegts temp2 2> /dev/null & \
ffmpeg -f mpegts -i "concat:temp1|temp2" -c copy -bsf:a aac_adtstoasc output.mp4 - concat_mp4.sh
- #!/bin/bash
MIN_ARGS=2
if (( $# < $MIN_ARGS ))
then
cat <<EOF
Usage: `basename $0` <input1> <input2> <input3> ... <output>
Concatenate several MP4 files into a single one.
EOF
exit 1
fi
#TMP=/tmp
TMP=$(mktemp -d)
# get all the command line parameters, except for the last one, which is output
output=${@:$#:1}
len=$(($#-1))
inputs=${@:1:$len}
function get_metadata {
video_path=$1
echo "[$(basename $0).get_metadata] $video_path"
info=$(ffprobe -i $video_path -show_format -v quiet)
metadata_str=$(echo "$info" | awk 'BEGIN {ORS=" ";FS="="} /^TAG:/ {gsub(/TAG:/,""); print "-metadata " $1 "=\"" $2 "\""}')
echo "[$(basename $0).get_metadata] metadata_str: $metadata_str"
return
}
function finish {
# remove created fifos
rm -f $TMP/mcs_*
}
# call finish function when exiting
trap finish EXIT
echo "[$(basename $0)]"
# remove all previous tmp fifos (if exist)
rm -f $TMP/mcs_*
i=0
for f in $inputs
do
echo "[$(basename $0)] input $i: $f"
if [[ -s $f ]]
then
# create fifo
fifo_name=$TMP/mcs_$i
mkfifo $fifo_name
# copy to fifo
ffmpeg -i $f -y -c copy -bsf:v h264_mp4toannexb -f mpegts $fifo_name 2> /dev/null &
if (( i==0 ))
then
concat_str="concat:${fifo_name}"
# get metadata (e.g. creation date) only from the first file
get_metadata $f
else
concat_str="${concat_str}|${fifo_name}"
fi
else
echo "[$(basename $0)] WARNING: $f is not a valid input"
fi
(( i++ ))
done
echo "[$(basename $0)] output: $output"
if [[ "$concat_str" ]]
then
echo "[$(basename $0)] concat_str: $concat_str"
eval ffmpeg -y -f mpegts -i \"${concat_str}\" -c copy -bsf:a aac_adtstoasc $metadata_str $output
else
echo "[$(basename $0)] ERROR: no valid input files found"
exit 1
fi
exit 0
- #!/bin/bash
- mkfifo temp1 temp2
- directly concatenable formats
- Using
an
external script
- mmcat.sh
- Concat
filter
- timestamps
- DTS: decoding time stamp
- PTS: presentation time stamp
- Info
- Sincronització / Synchronisation
- Tutorial 05: Synching Video (An ffmpeg and SDL Tutorial)
- [FFmpeg-user] Fixing bad DTS/PTS without reencode
- Recalculate PTS/DTS
- setpts
- multiple
input
overlay in 2x2 grid (wiki)
- setpts=PTS-STARTPTS
- multiple
input
overlay in 2x2 grid (wiki)
- -copyts
- -copytb mode
- -fflags genpts
- delay subtitles
- regenerate timestamps
- FFmpeg: recreate timestamps without reencoding
- video
-vsyncdrop- "As passthrough but destroys all timestamps, making the muxer generate fresh timestamps based on frame-rate."
- to reset "
start: xxxx.xxxxxx" to "start: 0.000000" (maybe you need to deal with video-only file?) ffmpeg -i toto_start_xxx.video.mp4 -c copy toto_start_0.video.mp4
- audio
- -async ...
- Problemes / Problems
- info output is written to stderr (stdout is used for media
data)
- ffmpeg -i toto.mp2 toto.aac 2>toto.log
- HOWTOs
- Pipe
- FFmpeg Howto (FFMedia Broadcast)
- Tutoriales FFmpeg para principiantes
- Using ffmpeg to manipulate audio and video files
- HOWTO: Fix Aspect ratio of Divx,etc for DVD with ffmpeg
- FFmpeg
and 'Fist of Fury'
- cropfilter: cropdetect
- subtitles
- DVD track extraction: tccat
- ...
- Encode
(el bocabyte vascodilatador)
- DVD track extraction: mplayer, vlc
- Shanes
video
encoding notes
- ..
- Frame-rate
- tbn, tbc, tbr
- -r ...
- -vsync
-
value
string
description
-1
auto
"Chooses between 1 and 2 depending on muxer capabilities. This is the default method." 0
passthrough "Each frame is passed with its timestamp from the demuxer to the muxer."
1
cfr
"Frames will be duplicated and dropped to achieve exactly the requested constant frame rate."
2
vfr
"Frames are passed through with their timestamp or dropped so as to prevent 2 frames from having the same timestamp."
drop
"As passthrough but destroys all timestamps, making the muxer generate fresh timestamps based on frame-rate."
- ffmpeg: -vsync 0 versus -r (frame rate)
-
- Create a video by fragmenting the provided one
into fragments of different framerates:
- vfr.sh
- #!/bin/bash
video_path=$1
video_basename=$(basename $video_path)
video_name=${video_basename%.*}
# video
video_coder=h264
video_bitrate=500k
# audio
audio_coder=aac
audio_bitrate=128k
# get duration
duration=$(ffprobe -i $video_path -show_format -v quiet | sed -n 's/duration=//p')
# keep only numbers and decimal point (discard N/A):
duration=$(echo $duration | tr -d -c 0-9.)
echo "duration of ${video_path}: ${duration}"
fps_array=(30 5 20 10 30 50 4 20 30 5 20 10 30 50 4 20 30 5 20 10 30 50 4 20 30 5 20 10 30 50 4 20 30 5 20 10 30 50 4 20 30 5 20 10 30 50 4 20 30 5 20 10 30 50 4 20 )
number_fragments=${#fps_array[@]}
echo "number_fragments: ${number_fragments}"
fragment_duration=$(echo ${duration}/${number_fragments} | bc )
echo "fragment_duration: ${fragment_duration}"
start=0
for fps in ${fps_array[@]}
do
echo "${fps} fps"
echo " start: $start"
output_filename=$(printf "%s_%05d_%sfps.mp4" ${video_name} ${start} ${fps})
echo " output_filename: ${output_filename}"
ffmpeg -y -ss ${start} -t ${fragment_duration} -i ${video_path} -vsync cfr -vf fps=${fps} -c:v ${video_coder} -b:v ${video_bitrate} -c:a ${audio_coder} -b:a ${audio_bitrate} ${output_filename}
start=$(( start + fragment_duration ))
done
# final concatenation
final_output_filename="${video_name}_vfr.mp4"
script=concat_mp4.sh
if which ${script}
then
echo "script ${script} found"
concat_mp4.sh -o ${final_output_filename} ${video_name}*fps.mp4
else
echo "script ${script} not found"
fi
exit 0
- #!/bin/bash
- vfr.sh
- Bit-rate
-
creation
info
OpenCaster
ffmpeg
x264
ffmpeg/ffplay
esvideompeg2info
-b:v 2M
bitrate from measured size (formula has rounds): 1997900.000000bps
-bt:v 200k
-minrate 2M
VBV: restrictions for hw decoder (e.g. DVB stb)
-maxrate 2M
--vbv-maxrate 2000
max. 2000 kb/s
Sequence header: bitrate: 2.00Mbs
-bufsize 1835008
--vbv-bufsize 1835
Sequence header: vbv buffer size: 112
-rc_init_occupancy 0 --vbv-init 0
- DVB bit-rates
- VBV in OpenCaster
- CBR in TS
- Bit-rate in H.264
- TS bit-rate
- OpenCaster
- tspcrmeasure muxed_13271000_stamped.ts 13000000
- OpenCaster
- Video Buffer Verifier (VBV)
- Compressió, multiplexació estadística i serveis avançats sobre TV digital terrestre (Rubén Santiso Puigarnau)
- Variable bit rates with
?-vb? and ?minrate?/?maxrate? settings in FFmpeg
- CBR: -b:v 3500K -minrate 3500K
- streaming: -maxrate 3500K -bufsize 1000K
- Limiting the output bitrate (ffmpeg wiki)
- EncodingForStreamingSites (ffmpeg wiki)
- To reduce
bit-rate (two passes: [Ffmpeg-user]
H.264
ignored bitrate) (-y: overwrite output files):
- ffmpeg -y -pass 1 ... -b 160k -qmax 50 ...
- ffmpeg -y -pass 2 ... -b 160k -qmax 50 ...
-
- H.264 video (using x264)
(see also libav
with libx264)
- FFmpeg and H.264 Encoding Guide
- FFmpeg x264 encoding guide
- Acceleració per maquinari / Hardware acceleration
- Linuxencoding
- X264 Encoding Suggestions
- X264 settings
- x264
FFmpeg
Options Guide (options for x264 and ffmpeg)
- default gop size: 250
- -g 250
- -keyint_min ... (will be truncated to gop_size/2 +1)
- -sc_threshold ... (scene cut detection)
- default gop size: 250
- GOP
- Key frame help
- Re:
changing
bitrate on the fly
- -sc_threshold ...
- fixed gop size (e.g. 12 frames)
- ... -g 12 -keyint_min 12 -sc_threshold 0 ...
- additional key frames:
- X264 Settings (MeWiki)
- High level settings
- Low level settings
- ... -x264opts "vbv-maxrate=2000:vbv-bufsize=200:slice-max-size=1500" ...
- x264
FFmpeg
Options Guide (options for x264 and ffmpeg)
- Baixa latència / Low latency
- Quality: CRF
Constant Rate Factor
-
crf
subjective quality
0
lossless
23
good
35
acceptable
51
terrible
-
- H.264
bit-rate
control
- A qualitative overview of x264's ratecontrol methods, By Loren Merritt
- VBV
- A qualitative overview of x264's ratecontrol methods
- X264 Settings
- Ratecontrol
- nal-hrd (Hypothetical Reference Decoder)
- x264 FFmpeg Options Guide
- -vpre
(use "-preset"
(quality related) and "-profile:v"
(profile constraints) instead)
- /usr/share/ffmpeg/libx264-{ipod320,ipod640}.ffpreset (git code) (those related to x264 have been replaced)
- libx264 (ffmpeg documentation)
- x264 ? FFmpeg Preset Choices (FFmpeg x264 encoding guide) (FFmpeg x264 encoding guide)
- -vpre quality_related -vpre profile_constraints:
-
quality related
profile constraints
- ultrafast
- superfast
- veryfast
- faster
- fast
- medium
- slow
- slower
- veryslow
- placebo
- baseline
- main
- ipod320
- ipod640
- Problems
- "maybe incorrect parameters such as bit_rate,
rate, width or height"
- Solution: use -vpre option
- when using -vpre ipodxxx, you must prepend a -vpre {slow...} option
- "maybe incorrect parameters such as bit_rate,
rate, width or height"
- Transport
stream
(TS) (see also libav
TS)
- Creating DVB-T compatible MPEG2 streams using FFMPEG
- Generar un TS (Transport Stream) valido para DVB-T y DVB-T2 (TDT) (-mpegts...)
- Encoding H.264 CBR videos with FFmpeg
- -streamid ...
- Mux bitrate (specified in bits per second) (Bitrate calculator)
- mpegts
encoding options:
- -mpegts_original_network_id
- -mpegts_transport_stream_id
- -mpegts_service_id
- -mpegts_pmt_start_pid
- -mpegts_start_pid
- -metadata service_provider="nom proveidor"
- -metadata service_name="nom servei"
- Program (service)
selection:
- Stream specifiers (bug)
- Convert MPEG2TS to H264 ES
- ffprobe
- ffprobe -v 0 -show_frames -select_streams v:1 794.ts
- play video (-vst) #0:11 and audio (-ast) #0:12
- ffplay -i 794.ts -vst 11 -ast 12
- from the first (0:) input file, select "Program
51" (:p:51:) video (:0), without transcoding (-c:v
copy) into H.264 elementary stream (input video
already was H.264):
- ffmpeg -i input.ts -map 0:p:51:0 -c:v copy output.h264
- select "program 802" video and two audios from
input file and generate a TS:
- ffmpeg -i 794.ts -map 0:p:802:0 -map 0:p:802:1 -map 0:p:802:2 -c:v copy -c:a copy /tmp/c33.ts
- or directly select the stream id given by ffmpeg
-i:
- ffmpeg -i 794.ts
- Program 803
Metadata:
service_name : ?3/24
service_provider: ?TVC
Stream #0:26[0x83]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p(tv), 720x576 [SAR 64:45 DAR 16:9], max. 15000 kb/s, 25 fps, 25 tbr, 90k tbn, 50 tbc
Stream #0:27[0x84](cat): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, stereo, s16p, 158 kb/s
Stream #0:25[0x86](v.o): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, mono, s16p, 94 kb/s
[...]
- Program 803
- ffmpeg -i 794.ts -map 0:26 -c:v copy /tmp/324.m2v
- ffmpeg -i 794.ts
- select by pid:
- get the PIDs:
- ffmpeg -i
- nput #0, mpegts, from
'tvc_794_20140821_1709.ts':
Duration: 13:26:44.25, start: 758.128889, bitrate: 26 kb/s
Program 801
Metadata:
service_name : ?TV3
service_provider: ?TVC
Stream #0:2[0x6f]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p(tv), 720x576 [SAR 64:45 DAR 16:9], max. 15000 kb/s, 25 fps, 25 tbr, 90k tbn, 50 tbc
Stream #0:7[0x70](cat): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, stereo, s16p, 158 kb/s
...
- nput #0, mpegts, from
'tvc_794_20140821_1709.ts':
- ffmpeg -i
- ff
- ffmpeg -f mpegts -i udp://host:port -c:a:i:2300 aac -b:a 384k -c:v copy -f flv rtmp://host/live/stream
- get the PIDs:
- Transcoding from TS with MPEG-2 audio and video to TS
with H.264 and MPEG-2 audio:
- ffmpeg -i tvc_hd_20121116-2348.ts -c:a mp2 -ab 64k -ar 48000 -ac 1 -async 1 -c:v libx264 -threads 4 -preset veryfast -profile:v main -b:v 200k -s 256x144 -f mpegts toto2.ts
- Creation of a TS from an ES
(obtained with ffmpeg -i 794.ts -map 0:p:802:0 -c:v copy
/tmp/c33.m2v):
- ffmpeg -fflags +genpts -i /tmp/c33.m2v -c:v copy /tmp/c33_m2v.ts
- MP4
(see also libav MP4)
- mov / mp4 / ismv
- Fragmentació /
Fragmentation
- How to convert an mpeg2 to a streamable or fragmented mp4?
- How to output
fragmented mp4 with ffmpeg?
- ffmpeg ... -movflags frag_keyframe+empty_moov ...
- H.264 + AAC
- ffmpeg -i ... -c:v libx264 -b:v 800k -g 25 -keyint_min 25 -sc_threshold 0 -c:a aac -b:a 128k toto.mp4
- MP4 files to be MPEG-DASHed with MP4Box,
and played
with
dash.js (which requires -frag-rap option in
MP4Box)
- two representations with an added key frame every
2 (n_forced*2)
seconds (see also libx264
GOP)
- ffmpeg -i sintel_trailer-720p.mp4 -c:v libx264 -b:v 500k -s 640x360 -aspect 16:9 -force_key_frames 'expr:gte(t,n_forced*2)' -c:a aac -strict -2 -b:a 128k sintel_trailer_640x360_500k.mp4
- ffmpeg -i sintel_trailer-720p.mp4 -c:v libx264 -b:v 1500k -s 1280x720 -aspect 16:9 -force_key_frames 'expr:gte(t,n_forced*2)' -c:a aac -strict -2 -b:a 128k sintel_trailer_1280x720_1500k.mp4
- simultaneous
two video representations and one audio
representation, read at its rate (-re):
- ffmpeg -re -i sintel_trailer-720p.mp4 \
-c:v libx264 -b:v 500k -s 640x360 -aspect 16:9 -keyint_min 24 -g 24 -r 24 -an sintel_trailer_640x360_500k.mp4 \
-c:v libx264 -b:v 1500k -s 1280x720 -aspect 16:9 -keyint_min 24 -g 24 -r 24 -an sintel_trailer_1280x720_1500k.mp4 \
-c:a aac -strict -2 -b:a 128k -vn sintel_trailer_audio_128k.mp4- Stream mapping:
Stream #0:0 -> #0:0 (h264 -> libx264)
Stream #0:0 -> #1:0 (h264 -> libx264)
Stream #0:1 -> #2:0 (aac -> aac)
- Stream mapping:
- ffmpeg -re -i sintel_trailer-720p.mp4 \
- two representations with an added key frame every
2 (n_forced*2)
seconds (see also libx264
GOP)
moov atom not found- Solució / Solution
- untrunc
- ponchio / untrunc
- Compilació / Compilation
- get the code from git
git clone https://github.com/ponchio/untrunc.gitcd untrunc
- compile with a compiled ffmpeg
(/usr/local)
g++ -o untrunc file.cpp main.cpp track.cpp atom.cpp mp4.cpp -L/usr/local/lib -I <path_to_src>/ffmpeg-3.4.2 -I/usr/local/include -lavformat -lavcodec -lavutil
- compile with a system ffmpeg
- ..
- get the code from git
- Ús / Usage
- ...
- untrunc fork
- anthwlock / untrunc
- Compilació / Compilation
- get the code from git
git clone https://github.com/anthwlock/untrunc.git untrunc_anthwlockcd untrunc_anthwlock
- compile
make FF_VER=4.2.3
- ...
- get the code from git
- untrunc
- Solució / Solution
- FLV
- ffmpeg -re -f lavfi -i "testsrc=size=864x480, format=pix_fmts=yuv420p" -f lavfi -i "sine=frequency=5000:sample_rate=44100, aformat=channel_layouts=stereo" -c:v h264 -b:v 600k -profile:v main -preset ultrafast -c:a aac -strict -2 -b:a 64k -f flv toto.flv
- RTMP
- Test
Streaming
With FFMPEG
- ffmpeg -re -f lavfi -i
"testsrc=size=1280x720:rate=24" -f lavfi -i
sine=f=220:b=1 -af "volume=0.1" -codec:a aac
-strict -2 -codec:v libx264 -preset ultrafast
-profile:v main -tune zerolatency -pix_fmt
yuv420p -g 48 -f flv rtmp://...
- Problems
- [flv @ 0x23df540] Packets are not in
the proper order with respect to DTS
av_interleaved_write_frame(): Invalid argument
[flv @ 0x23df540] Failed to update header with correct duration.
[flv @ 0x23df540] Failed to update header with correct filesize.- Solution: update ffmpeg
- problem is present with:
- libavformat 56. 4.101
- it is not present with:
- libavformat 56. 25.101
- problem is present with:
- Solution: update ffmpeg
- [flv @ 0x23df540] Packets are not in
the proper order with respect to DTS
- Problems
- ffmpeg -re -f lavfi -i
"testsrc=size=1280x720:rate=24" -f lavfi -i
sine=f=220:b=1 -af "volume=0.1" -codec:a aac
-strict -2 -codec:v libx264 -preset ultrafast
-profile:v main -tune zerolatency -pix_fmt
yuv420p -g 48 -f flv rtmp://...
- Test
Streaming
With FFMPEG
- Specific formats
- Android
video
- Video
encoding
recommendations (Android Developers)
lower quality
higher quality
video
codec
H.264 baseline profile
H.264 baseline profile
resolution
176x144
480x360
frame rate
12 fps
30 fps
bitrate
56 Kbps
500 Kbps
audio
codec
AAC-LC
AAC-LC
channels
1
2
bitrate
24 Kbps
128 Kbps
- Converting Videos For The Android T-Mobile G1 Phone With Linux
- Encoding Video for Android (Broken links)
- PAL 16:9
to 480x320 ("-map 0:2" to use 2nd audio track) (DVD track extraction):
- ffmpeg -threads 4 [-ss seconds] -i toto.vob -map 0:0 -map 0:2 -s 480x270 -padtop 24 -padbottom 26 -vcodec libx264 -vpre superfast -vpre baseline -b 480k -acodec libvo_aacenc -ab 128k -async 2 -sameq -aspect 3:2 toto_android.mp4
- HD 1080i
to 480x320 ("-ac 2" to downmix to 2 audio channels):
- ffmpeg -threads 4 -i toto.MTS -s 480x270 -padtop 24 -padbottom 26 -vcodec libx264 -vpre superfast -vpre baseline -b 480k -acodec libvo_aacenc -ab 128k -ac 2 -r 25 -sameq -aspect 3:2 toto_android.mp4
- ffmpeg -aspect 3:2 -i inputfilename.ext -aspect 3:2 -s 480x320 -vcodec h264 -b 480k -r 23.976 -acodec aac -ab 96k -sameq -pass 1 outputfilename.mp4
- ffmpeg -aspect 4:3 -i inputfilename.ext -aspect 3:2 -s 400x300 -vcodec h264 -b 480k -r 23.976 -acodec aac -ab 96k -padtop 10 -padbottom 10 -padleft 40 -padright 40 -sameq -pass 1 outputfilename.mp4
- ffmpeg -aspect 16:9 -i inputfilename.ext -aspect 3:2 -s 480x270 -vcodec h264 -b 480k -r 23.976 -acodec aac -ab 96k -padtop 24 -padbottom 26 -sameq -pass 1 outputfilename.mp4
- Video
encoding
recommendations (Android Developers)
- HbbTV
video
- SD
- ffmpeg -i toto.mp4 -s 720x576 -aspect 16:9 -r 25 -vcodec libx264 -vpre fast -sameq toto-576.mp4
- HD
- ...
- SD
- TomPlayer
video
- ffmpeg -i toto.xxx -vcodec mpeg4 -b 700k -s 480x272 toto.avi
- iPhone video
- iPhone (Videolan)
- iPhone HTTP streaming
- video
transcoding recipes with ffmpeg
- ffmpeg -i input_file -f mp4 -vcodec mpeg4 -maxrate 1000k -b 700k -qmin 3 -qmax 5 -bufsize 4096k -g 300 -acodec libfaac -ab 192k -s 480x320 -aspect 4:3 output_file.mp4
- x264
encoding
guide
- --level 30 --profile baseline --vbv-bufsize 10000 --vbv-maxrate 10000
- iPod video
- iPod (Videolan)
- iPod
video
guide
- 1.5Mbit/s 640×480 H.264 videos
- ffmpeg -i INPUT -acodec libfaac -ab 128k -s WIDTHxHEIGHT -vcodec libx264 -vpre slow -vpre ipod640 -b BIT_RATE -aspect WIDTH:HEIGHT -threads 0 -f ipod OUTPUT.mp4
- 768kbit/s 320×240 H.264 videos
- ffmpeg -i INPUT -acodec libfaac -ab 128k -s WIDTHxHEIGHT -vcodec libx264 -vpre slow -vpre ipod320 -b BIT_RATE -metdata "title=SOME_TITLE" -threads 0 -f ipod OUTPUT.mp4
- 1.5Mbit/s 640×480 H.264 videos
- commandline ffmpeg & x264 settings for Ipod 5.5
- (*): ffmpeg -i input.mp4 -f mpegts -acodec libmp3lame -ar 48000 -ab 64k -s 320x240 -vcodec libx264 -b 96k -flags +loop -cmp +chroma -partitions +parti4x4+partp8x8+partb8x8 -subq 5 -trellis 1 -refs 1 -coder 0 -me_range 16 -keyint_min 25 -sc_threshold 40 -i_qfactor 0.71 -bt 200k -maxrate 96k -bufsize 96k -rc_eq 'blurCplx^(1-qComp)' -qcomp 0.6 -qmin 10 -qmax 51 -qdiff 4 -level 30 -aspect 320:240 -g 30 -async 2 output.ts
- ffmpeg -i <in file> -f mpegts -acodec
libmp3lame -ar 48000 -ab 64k -vcodec libx264 -s
320x240 -b 96k -bt 200k -maxrate 96k -bufsize 96k
-rc_eq 'blurCplx^(1-qComp)' -level 30 -aspect
320:240 -async
2 -vpre mobile <output file>
- libx264-mobile.ffpreset:
- flags=+loop
cmp=+chroma
partitions=+parti4×4+partp8×8+partb8×8
subq=5
trellis=1
refs=1
coder=0
me_range=16
keyint_min=25
sc_threshold=40
i_qfactor=0.71
qcomp=0.6
qmin=10
qmax=51
qdiff=4
g=30
- flags=+loop
- libx264-mobile.ffpreset:
- x264
encoding
guide
- --level 1.3 --no-cabac --vbv-bufsize 768 --vbv-maxrate 768
- PS3 video
- Conversion for Portivity
(60s starting at time 50)
- ffmpeg -ss 50 -t 60 -i sp.mxf -vcodec libx264 -s 320x240 -b 160k -r 12.5 -acodec libfaac -ab 32k sp.mp4
- [ffmpeg -ss 50 -t 60 -i sp.mxf -vcodec h264 -s 320x240 -b 160k -r 12.5 -acodec aac -ab 32k sp.mp4]
- MP4Box -hint sp.mp4 (to be streamed by Darwin server)
- To Samsung SVI (no
funciona...):
- ffmpeg -i toto.mp4 -f avi -s 208x176 -r 15 -vcodec xvid toto.svi
- Android
video
- Just copy the audio ES
- ffmpeg -acodec copy ...
- General (ts -> h264/MP4) (plf needed):
- ffmpeg -i toto.ts -vcodec {h264,libx264} -s 320x240 -r 15 -b 1000 -f mp4 toto.mp4
- Elementary streams
- From .m2v
(elementary stream) to .h264 (elementary stream):
- ffmpeg -i toto.m2v [-f h264] toto.h264
- From .mp2
(elementary stream) to AAC (elementary stream) (plf needed):
- ffmpeg -i toto.mp2 toto.aac
- From .m2v
(elementary stream) to .h264 (elementary stream):
- YUV
- From .yuv video
file (-s must be the first parameter):
- ffmpeg -s 704x576 -i toto_704x576.yuv -f mp4 -vcodec {h264,libx264} toto.mp4
- ffmpeg -s 720x576 -i 576i25_shields_ter.yuv -target pal-dvd -aspect 16:9 576i25_shields_ter.vob
- ffmpeg -s 1280x720 -r 50 -i 720p50_shields_ter.yuv -target pal-dvd -aspect 16:9 720p50_shields_ter.vob
- To .yuv video
file:
- ffmpeg -i toto.mp4 toto.yuv
- From .yuv video
file (-s must be the first parameter):
- Video codecs
- MPEG-4
SP (simple profile):
- ffmpeg ... -profile 0...
- To force MPEG-4 ASP (advanced
simple profile):
- ffmpeg ... -profile 15...
- To specify H.264 level:
- ffmpeg ... -[v]level 30 ... (level 3.0)
- To specify video profile (prefer -profile:v instead of -vprofile):
-
-profile:v MPEG-4 H.264 (libx264) 0, 16 Simple Baseline 1, 17 Simple Scalable Baseline 2 Core Baseline 3 Unknown Profile(49)
4 Unknown Profile(65
5 Scalable Texture
6 Simple Face Anim
7 Basic Anim Text
8 Hybrid
9 Adv RT Simple
10 Core Scalable
11 Adv Coding Efficieny
12 Adv Core Profile
13 Adv Scalable Texture
14 Simple Studio
15 Adv Simple
- AVCHD
to AVI (AVS
forum)
- FFmpeg Tutorial: Convert MTS/M2TS(AVCHD) to AVI
- % xport -h sample.m2ts 1 1 1
% mv bits0001.mpa sample.ac3
% mkfifo pipe.yuv
% ldecod -i bits0001.mpv -o pipe.yuv &
% ffmpeg -r 29.97 -s 1440x1080 -i pipe.yuv -i /tmp/sample.ac3 -acodec copy -vcodec mpeg4 -aspect 16:9 -b 15000k sample.avi
% rm pipe.yuv
- GIF animat / Animated GIF
- MPEG-4
SP (simple profile):
- I-frames only
- fmpeg ... -intra ...
- B-frames
ffmpeg ... -bf number ...- No B-frames
ffmpeg ... -bf 0 ...
- image
-> video
- Timelapse / Stop
motion
- Subsample image file names
- cp ranges
- From several
jpeg or png files:
- Create a video slideshow from images
- Reproducció / Play
- ffplay -framerate 50 -start_number 150 -i DSC%05d.JPG
- Creació de vídeo / Video creation
- to 1280x720@8fps:
- ffmpeg -y -start_number 767 -r 8 -i DSC%05d.JPG -vf 'pad=max(iw\,ih*16/9):ow*9/16:(ow-iw)/2:(oh-ih)/2, scale=1280:-1' -c:v libx264 -f mp4 toto_8fps.mp4
- ffmpeg -start_number 150 -i DSC%05d.JPG -c:v libx264 -f mp4 toto.mp4
- ffmpeg -r 5 -start_number 150 -i DSC%05d.JPG -vf scale=1280:-1 -c:v libx264 -f mp4 toto.mp4
- ffmpeg -r 15 -start_number 150 -i toto%02d.png -f mp4 -vcodec {h264,libx264} toto.mp4
- to 1280x720@8fps:
- From all png files in a dir (globbing):
- ffmpeg ... -pattern_type glob -i 'foo-*.jpeg' ...
- Smooth
- ffmpeg convert a series of images to video - with crossfade or any other transition between every two frames
- blend filter
- ffmpeg -i {input} -vf "tblend=average,framestep=2,setpts=0.50*PTS" -r {srcfps} -{encoding parameters} {output}
- Butterflow
- From video
- ffmpeg -i input.mp4 -filter:v "setpts=0.5*PTS" -an output.mp4
- Make a Video File from a Single
Frame:
- ffmpeg -loop_input -vframes <number_of_frames> -i <input_file> <output_file>
- Carta
d'ajust / Test card
- ffmpeg -loop 1 -i toto.png -f lavfi -i "sine=frequency=440:sample_rate=48000,aformat=channel_layouts=stereo" -vf "setdar=dar=16/9" -t 10 -pix_fmt yuv420p -c:v h264 -c:a aac toto_169.mp4
- ffmpeg -loop 1 -i PM5544_with_non-PAL_signals.png -f lavfi -i "sine=frequency=440:sample_rate=48000,aformat=channel_layouts=stereo" -filter:v "scale=w=720:h=576,setdar=dar=4/3" -t 10 -pix_fmt yuv420p -c:v h264 -c:a aac pm5544.mp4
- ffmpeg -loop 1 -i Philips_Pattern_pm5644.png -f lavfi -i "sine=frequency=440:sample_rate=48000,aformat=channel_layouts=stereo" -filter:v "scale=w=720:h=576,setdar=dar=16/9" -t 10 -pix_fmt yuv420p -c:v h264 -c:a aac pm5644.mp4
- ffmpeg -loop_input -vframes 250 -i bars_grey.bmp -vcodec mpeg2video bars_grey.mpeg
- From raw (c-generated)
rgb images:
- tvTestPattern > f.raw
- ffmpeg -f rawvideo -pix_fmt rgb24 -s 720x576 -aspect 4:3 -i f.raw -f s16le -ac 2 -i /dev/zero -shortest -target dvd -b 4000 -ab 128 -ildct -ilme -y interlace_test2.mpeg
- From raw UYVY 4:2:2
images:
- ffmpeg -loop_input -f rawvideo -pix_fmt uyvy422 -s
720x576 -aspect 4:3 -i mostra.uyvy -f s16le -ac 2 -i
<(dd if=/dev/zero bs=19200 count=1) -target
pal-dvd toto.mpeg
- ffplay toto.mpeg (durada infinita!) (de fet, es pot fer directament: ffplay -s 720x576 -f rawvideo -pix_fmt uyvy422 -i toto.uyvy)
- ffmpeg -loop_input -f rawvideo -pix_fmt uyvy422 -s
720x576 -aspect 4:3 -i mostra.uyvy -f s16le -ac 2 -i
<(dd if=/dev/zero bs=19200 count=1) -target
pal-dvd toto.mpeg
- mpeg animation
(eg for DVD menus) from a single image (*):
- ffmpeg -loop_input -t 1.0 -i stillframename -ar 48000 -f s16le -i <(dd if=/dev/zero bs=19200 count=1) -target pal-dvd outputmoviename
- Timelapse / Stop
motion
- video -> images
(thumbnails)
- WebP
- libwebp
- Compilació / Compilation
- Ús / Usage
- Convert MP4 file to animated WEBP file in ffmpeg CLI
ffmpeg -i input.mp4 -vcodec libwebp -filter:v fps=fps=2 -loop 1 -an -vsync 0 -s 128:72 output.webp
- Extract only (-vframes 1 or -t 00:00:00:01) the
5th frame of second 20 (-ss
00:00:20:05), from file into a jpeg image
(thumbnail) (*)
(-ss before -i is much faster):
- ffmpeg -ss 00:00:20:05 -i toto.mp4 -vframes 1 -f mjpeg toto.jpeg
- ffmpeg -ss 00:00:20:05 -i toto.mp4 -vframes 1 -f singlejpeg -s 150x100 toto.jpeg (maybe singlejpeg is not supported)
- ffmpegthumbnailer
- Extract all frames from file into several
jpeg
files (implicit image2
format):
- ffmpeg -i toto.mp4 toto%d.jpeg
- Extract one frame every 1 second (-r 1) to several jpeg images:
- Multiple
thumbnails:
- ffmpeg -ss $i*$interval -i intro.mov -vframes 1 -s 320x240 thumb_$i.jpg
- thumb.sh (allows output name of type
img_<position_in_seconds>s.jpeg)
- #!/bin/bash
input=$1
interval=$2
# output dimensions (aspect ratio will be preserved)
out_width=320
out_height=240
# video duration
duration=$(ffprobe -i $input -show_format -v quiet | sed -n 's/duration=//p')
duration_int=${duration%\.*}
echo "duration: $duration"
position=0
while (( position < duration_int ))
do
position_name=$(printf "%06d" $position)
echo $position_name
ffmpeg -v 0 -ss $position -i $input -vframes 1 -vf scale=w=${out_width}:h=${out_height}:force_original_aspect_ratio=decrease img_${position_name}s.jpeg
(( position=position+interval ))
done
- #!/bin/bash
- ffmpeg-php
- Generating Video Thumbnail Storyboards
- Making thumbnails for movies with ffmpeg
- WebP
- Crop
- part central del vídeo, amb relació 4:3
- ffplay sintel-1024-stereo.mp4 -vf 'crop=ih*4/3:ih'
- part central del vídeo, amb relació 4:3
- Relació
d'aspecte / Aspect ratio
-
input image output image satisfy only one destination dimension
satisfy both destination dimensions
width height inscribe excribe adapted pad
- -vf "scale=${width}:-1"
- # force height to be multiple of 2
-vf "scale=${width}:-2"
- -vf "scale=-1:${height}"
- # force width to be multiple of 2
-vf "scale=-2:${width}"
- -vf "scale=${width}:${height}:force_original_aspect_ratio=decrease"
- # force dimensions to be multiple of 2
(using pad)
-vf "scale=${width}:${height}:force_original_aspect_ratio=decrease, pad='iw+mod(iw\,2)':'ih+mod(ih\,2)'" - # force dimensions to be multiple of 2 (a
recent version of ffmpeg is needed)
-vf "scale=${width}:${height}:force_original_aspect_ratio=decrease:force_divisible_by=2"
-vf "scale=${width}:${height}:force_original_aspect_ratio=increase" -vf "scale=${width}:${height}" - this is the most generic solution to automatically add pillarbox or letterbox
- first, we calculate the video region
- min is used to allow pillarbox or letterbox
- dar must be taken into account
- output dar must be calculated offline, because ffmpeg does not do it (only for input dar)
- ...
- then we put the video on centre and pad it
with black
- pad=${out_width}:${out_height}:-1:-1
- out_width=720
out_height=576
out_dar="4:3"
out_dar_bar=${out_dar//:/\/}
out_dar_float=$(echo "${out_dar_bar}" | bc -l)
ffmpeg -y -v verbose -i ${input} -vf "scale=min(${out_width}\,trunc(${out_width}*dar/${out_dar_float}+0.5)):min(${out_height}\,trunc(${out_height}*${out_dar_float}/dar+0.5)), pad=${out_width}:${out_height}:-1:-1" ${output}
PAL
720x576, 4:3768x576 4:3 -vf "scale=768:-1"
-> 768x614, 4:3
(ok, but too high)
-vf "scale=-1:576"
-> 720x576, 4:3 (ok)
-vf "scale=768:576:force_original_aspect_ratio=decrease"
-> 720x576, 4:3 (ok)
-vf "scale=768:576:force_original_aspect_ratio=increase"
-> 768x614, 4:3 (ok, but too high)
-vf "scale=768:576"
-> 768x576, 4:3 (ok)
1024x576
16:9
PAL anamorphic
720x576, 16:9
- -vf "scale=540:576,
pad=720:576:-1:-1"
-> 720x576, 16:9
1280x720 16:9 -vf "scale=1280:-1"
-> 1280x1024, 4:3
(ok, but too high)
-vf "scale=-1:720"
-> 900x720, 4:3 (ok)
-vf "scale=1280:720:force_original_aspect_ratio=decrease"
-> 900x720, 4:3 (ok)-vf "scale=1280:720:force_original_aspect_ratio=increase"
-> 1280*1024, 4:3 (ok, but too high)
PAL anamorphic
720x576, 16:91024x576 16:9
-vf "scale=1024:576:force_original_aspect_ratio=decrease"
-> 720x576, 16:9 (ok, still anamorphic)
-vf "scale=1024:576"
-> 1024x576, 16:9 (ok)
PAL
720x576, 4:3
1280x720 16:9 -vf "scale=1280:-1"
-> 1280x1024, 4:3
(ok, but too high)-vf "scale=-1:720"
-> 900x720, 16:9 (ok, still anamorphic)-vf "scale=1280:720:force_original_aspect_ratio=decrease"
-> 900x720, 16:9 (ok, still anamorphic)
1280x720, 16:9 PAL
720x576, 4:3
1280x720, 16:9 PAL anamorphic
720x576, 16:9
- Variables
-
input description output in_w
iwwidth out_w
owin_h
ihheight out_h
oha = (iw/ih) relation
sar sample aspect ratio
dar = (iw/ih)*sar display aspect ratio
-
- vegeu pad amb libav / see simplified pad with libav
- scale
filter preserving aspect ratio:
- "The scale filter forces the output display aspect ratio to be the same of the input, by changing the output sample aspect ratio."
- force width:
- ffmpeg ... -vf scale=320:-1...
- force width, but height has to be multiple of n=2
(n can be any value):
- ffmpeg ... -vf scale=320:-2...
- inscribe into a maximum rectangle (*):
- Resizing videos with ffmpeg/avconv to fit into static sized player
- ffmpeg ... -vf "scale=w=854:h=480:force_original_aspect_ratio=decrease"
...
- to force multiple of n (a recent version
of ffmpeg is needed):
- ... -vf "scale=320:h=180:force_original_aspect_ratio=decrease:force_divisible_by=2" ...
- to force multiple of 2 using pad:
- ... -vf "scale=320:h=180:force_original_aspect_ratio=decrease, pad='iw+mod(iw\,2)':'ih+mod(ih\,2)'" ...
- to force output aspect (
setdar=dar) to the same as input aspect (=dar):- ... setdar=dar=dar:max=16 ...
- complete example:
- ffmpeg -y -f lavfi -i
"testsrc=duration=10:size=720x576,
format=pix_fmts=yuv420p" -f lavfi -i
anoisesrc=duration=10:colour=pink:amplitude=0.5
-ac 2 -aspect 4:3 vídeo_43.mp4
ffmpeg -i vídeo_43.mp4 -vf "scale=320:h=180:force_original_aspect_ratio=decrease, pad='iw+mod(iw\,2)':'ih+mod(ih\,2)', setdar=dar=dar:max=16" -c:a copy video_43_petit.mp4
- ffmpeg -y -f lavfi -i
"testsrc=duration=10:size=720x576,
format=pix_fmts=yuv420p" -f lavfi -i
anoisesrc=duration=10:colour=pink:amplitude=0.5
-ac 2 -aspect 4:3 vídeo_43.mp4
- to force multiple of n (a recent version
of ffmpeg is needed):
- ffmpeg -y -f lavfi -i
"testsrc=duration=10:size=720x576,
format=pix_fmts=yuv420p" -f lavfi -i
anoisesrc=duration=10:colour=pink:amplitude=0.5
-ac 2 -aspect 16:9 vídeo_169.mp4
ffmpeg -i vídeo_169.mp4 -vf "scale=320:h=180:force_original_aspect_ratio=decrease, pad='iw+mod(iw\,2)':'ih+mod(ih\,2)', setdar=dar=dar:max=16" -c:a copy video_169_petit.mp4 - ffmpeg ... -vf scale='ceil(min(1\,min(854/iw\,480/ih))*iw):-2' ...
- ffmpeg ... -vf scale='-2:ceil(min(1\,min(854/iw\,480/ih))*ih)' ...
- excribe around a minimum rectangle:
- ffmpeg ... -vf scale=w=854:h=480:force_original_aspect_ratio=increase ...
- # pad
the input to get a final w/h ratio of 16:9
pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2" - resize (1920x1080 16:9 -> 720x576 16:9
anamorphic) and transcode (H.264, AAC):
- avconv -threads auto -i toto.MTS -filter:v "scale=w=720:h=576,setdar=dar=16/9" -strict experimental toto.mp4
- resize from 1280x544 (2,35)
- to SD 4:3 (with letterbox)
- ffplay -i sintel-1280-surround.mp4 -vf 'pad=width=iw:height=iw*3/4:x=0:y=(oh-ih)/2, scale=w=720:h=576, setdar=dar=4/3'
- ffmpeg -i sintel-1280-surround.mp4 -vf 'pad=width=iw:height=iw*3/4:x=0:y=(oh-ih)/2, scale=w=720:h=576, setdar=dar=4/3' -c:v mpeg2video -an sintel_pad_4\:3_v2.mp4
- to HD 16:9 (with letterbox)
- ffplay -i sintel-1280-surround.mp4 -vf 'pad=width=iw:height=iw*9/16:x=0:y=(oh-ih)/2'
- to SD 16:9 anamorphic (with letterbox)
- ffplay -i sintel-1280-surround.mp4 -vf 'pad=width=iw:height=iw*9/16:x=0:y=(oh-ih)/2, scale=w=720:h=576, setdar=dar=16/9'
- ffmpeg -i sintel-1280-surround.mp4 -vf 'pad=width=iw:height=iw*9/16:x=0:y=(oh-ih)/2, scale=w=720:h=576, setdar=dar=16/9' -c:v libx264 -an sintel_pad_16:9.mp4
- to SD 4:3 (with letterbox)
- ...
- pad
video filter
- output sar is the same as input sar; dar may change according to this
- qualsevol relació d'aspecte d'entrada, cap a
sortida en 16:9
- ffplay sintel-1024-stereo.mp4 -vf 'pad=max(iw\,ih*16/9):ow*9/16:(ow-iw)/2:(oh-ih)/2'
- ...
- center
- ... pad=1280:720:-1:-1
- From anamorphic
(active: 720x576) to letterbox (ETSI
300
294) (active: 720x432 (720/16*9 *
4/3/(720/576)=432); horizontal bars: (576-432)/2=72
heigth each) (padcolor will usually be black 000000)
(sameq to have the same bitrate as the input):
- ffmpeg ... -s 720x432 -padtop 72 -padbottom 72 -padcolor 00ff00 -sameq -aspect 4:3 ...
- ...
4:3
14:9
16:9
720x576 768x576 720x576 896x576 720x576 1024x576 1280x720 1920x1080 3840x2160 dar size sar name create with ffmpeg 16:15 1:1 56:45 1:1 64:45 1:1 1:1 1:1 1:1 4:3 720x576 16:15 PAL -f lavfi -i "testsrc=duration=10:size=720x576, format=pix_fmts=yuv420p" -aspect 4:3 video_720x576_43.mp4
768x576 1:1
-f lavfi -i "testsrc=duration=10:size=768x576, format=pix_fmts=yuv420p" video_768x576_43.mp4
14:9 720x576 56:45 PAL 14:9 -f lavfi -i "testsrc=duration=10:size=720x576, format=pix_fmts=yuv420p" -aspect 14:9 video_720x576_149.mp4
896x576 1:1
-f lavfi -i "testsrc=duration=10:size=896x576, format=pix_fmts=yuv420p" video_896x576_149.mp4
16:9 720x576 64:45 PAL anamorphic -f lavfi -i "testsrc=duration=10:size=720x576, format=pix_fmts=yuv420p" -aspect 16:9 video_720x576_169.mp4
1024x576 1:1
-f lavfi -i "testsrc=duration=10:size=1024x576, format=pix_fmts=yuv420p" video_1024x576_169.mp4
1280x720 1:1 720 -f lavfi -i "testsrc=duration=10:size=1280x720, format=pix_fmts=yuv420p" video_1280x720_169.mp4
1920x1080 1:1 1080 -f lavfi -i "testsrc=duration=10:size=1920x1080, format=pix_fmts=yuv420p" video_1920x1080_169.mp4
3840x2160 1:1 4K, UHD -f lavfi -i "testsrc=duration=10:size=3840x2160, format=pix_fmts=yuv420p" video_3840x2160_169.mp4
-
4:3
16:9
720x576 720x576 4:3
720x576
-
- -vf "scale=540:576, pad=720:576:-1:-1"
- -s 540x576 -padleft 90 -padright 90 -aspect 16:9
768x576
-s 720x576
-aspect 4:314:9
720x576
-s 720x492
-padtop 42 -padbottom 42
-aspect 4:3
- -vf "scale=630:576, pad=720:576:-1:-1"
- -s 630x576 -padleft 44 -padright 46 -aspect 16:9
896x576
16:9
720x576
-s 720x432
-padtop 72 -padbottom 72
-aspect 4:3-
1024x576
- -vf "scale=720:576"
- -s 720x576 -aspect 16:9
-
NOTE: scale preserves dar; pad preserves sar
4:3 16:9 dar
sar 720x576 (out_sar=16/15) 720x576 (out_sar=64/45) 4:3 720x576 16:15 - - -vf "scale=540:576, pad=720:576:-1:-1"
768x576 1:1 - -vf "scale=720:576"
- -s 720x576 -aspect 4:3
14:9 720x576 56:45
- -vf "scale=..."
- -s 720x492 -padtop 42 -padbottom 42 -aspect 4:3
- -vf "scale=630:576, pad=720:576:-1:-1"
896x576 1:1 16:9 720x576 64:45 - -vf "..."
- -s 720x432 -padtop 72 -padbottom 72 -aspect 4:3
- 1024x576 1:1 - -vf "scale=720:576"
- -s 720x576 -aspect 16:9
- Parameter calculation:
- intermediate_width = in_width * (in_sar) /
(out_sar) * out_height / in_height = in_dar /
out_dar * out_width
- examples:
- 720 * (16/15) / (64/45) = 540
- 720 * (56/45) / (64/45) = 630
- examples:
- intermediate_height = in_height * (out_sar) /
(in_sar) * out_width / in_width = out_dar / in_dar *
out_height
- examples:
- ...
- examples:
- intermediate_width = in_width * (in_sar) /
(out_sar) * out_height / in_height = in_dar /
out_dar * out_width
- Càlculs / Calculations (aspecte.ods, aspecte.c):
- Input image
- Wi: given input image width
- Hi: given input image height
- DARi: given input display aspect ratio
- input anamorphic factor: Di = DARi / (Wi/Hi)
- Output image
- Wo: given output image width
- Ho: given output image height
- DARo: given output display aspect ratio
- output anamorphic factor: Do = DARo / (Wo/Ho)
- W'o: calculated output active image width
- H'o: calculated output active image height
- LR: calculated total number of pixels at sides (LR/2 at each side)
- TB: calculated total number of pixels at top and bottom (TB/2 at each zone)
- Pillar box (if calculated LR
gives >0)
- W'o=Wi * Di/Do * Ho/Hi
- H'o=Ho
- LR=Wo - W'o
- Letter box (if calculated TB
gives >0)
- W'o=Wo
- H'o=Hi * Do/Di * Wo/Wi
- TB=Ho - H'o
- Input image
-
- Rotació / Rotation
- Tall / Cut
- -t <duration>
- -to <position>
- Obté un tall de 10 segons a partir de la posició
00:02:00.0
ffmpeg -ss 00:02:00.0 -t 00:00:10.0 -i toto.mp4 -c copy tall_00:02:00.0-00:02:10.0.mp4ffmpeg -ss 00:02:00.0 -i toto.mp4 -t 00:00:10.0 -c copy tall_00:02:00.0-00:02:10.0.mp4
- Obté un tall que va dels 00:02:00.0 als 00:02:20.0
ffmpeg -ss 00:02:00.0 -to 00:02:20.0 -i toto.mp4 -c copy tall_00:02:00.0-00:02:20.0.mp4- If you put -to option after -i option, you will get a file with duration 00:02:20.0, as -to position will be referred to the output
- Extracció
/ Extraction
- Transport stream
- Extract only video
track
- ffmpeg -i toto.vob -an -vcodec copy toto.m2v
- Extract only audio
(vn: no video) track:
- ffmpeg -i toto.mxf -vn -acodec copy toto.mp2
- ffmpeg -i toto.vob -map 0:3 -vn -acodec copy toto.wav
- Extract only subtitles track:
- ffmpeg -i toto.mov -vn -an -scodec copy toto.srt
- ffmpeg -i toto.mov -map s:0 -vn -an -c:s copy toto2.srt
- extract the second subtitles track from program
801 in a TS:
- ffmpeg -i toto.ts ...
- extract the cat dvb subtitles (in bold) from a SP
TS file:
- ffmpeg -i 120208_2318\(33\).ts
Program 802
Metadata:
service_name : ?33
service_provider: ?TVC
Stream #0:0[0x79]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p, 720x576 [SAR 64:45 DAR 16:9], 15000 kb/s, 26.94 fps, 25 tbr, 90k tbn, 50 tbc
Stream #0:1[0x7a](cat): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, stereo, s16, 160 kb/s
Stream #0:2[0x7c](v.o): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, mono, s16, 96 kb/s
Stream #0:3[0x7e](a.d): Audio: mp2 ([3][0][0][0] / 0x0003), 48000 Hz, mono, s16, 96 kb/s
Stream #0:4[0x85](cat): Subtitle: dvb_teletext ([6][0][0][0] / 0x0006)
Stream #0:5[0x322](cat): Subtitle: dvb_subtitle ([6][0][0][0] / 0x0006)
Stream #0:6[0x32e](eng): Subtitle: dvb_subtitle ([6][0][0][0] / 0x0006)
Stream #0:7[0x386]: Unknown: none ([5][0][0][0] / 0x0005)
Stream #0:8[0x259]: Unknown: none ([12][0][0][0] / 0x000C) - ffmpeg -i 120208_2318\(33\).ts -map 0:5 -vn -an -scodec copy toto.srt
- ffmpeg -i 120208_2318\(33\).ts -map 0:5 -vn -an -scodec copy -f rawvideo dvbsub.dat (*)
- ffmpeg -i 794_20121105_0026.ts -map 0:8 -vn -an -scodec copy -copyinkf -f srt dvbsub.srt (*)
- ffmpeg -i 120208_2318\(33\).ts
- [Ffmpeg-user]
encode
multiple audio tracks to a single output file
- ffmpeg -i toto.vob -map 0:0 -map 0:1 -map 0:2 -acodec copy -vcodec copy toto_multi.vob -acodec copy -newaudio
- Multiangle
generation:
- ffmpeg -i toto1.m2v -i toto2.m2v -vcodec copy toto.vob -vcodec copy -newvideo
- Convert from VOB
to TS:
- ffmpeg -i toto.vob -sameq -acodec copy toto.ts
- Multiplex ES:
- ffmpeg -i toto.h264 -i toto.aac -vcodec copy -acodec copy toto.mp4
- Àudio / Audio
- Multicanal / Multichannel
- Manipulating audio channels
- Text to speech
flite -v -voice slt -t "left channel" FL.wav
- Dolby ProLogic II
- How to determine if an audio track is a Dolby Pro Logic II mixdown
- Create test 7.1, 5.1 and Pro Logic II audio/video files
ffmpeg -i toto_multichannel.mp4 -c:a flac -af "aresample=matrix_encoding=dplii" -ac 2 toto_dplii.flac
- Multicanal / Multichannel
- Sincronització
àudio/vídeo / Audio/video synchronisation
- aresample
async
(audio
filter)
- FFmpeg Resampler Documentation
- "Setting this to 1 will enable filling and
trimming"
ffmpeg ... -af aresample=async=1 ...
- Fixing out-of-sync audio and video with ffmpeg
-itsoffset
- aresample
async
(audio
filter)
- Subtítols / Subtitles
- list of supported subtitle codecs:
- ffmpeg -codecs | grep "^...S"
- extract
subtitles
- delay and hardcode (using -filter_complex)
subtitles:
- ffmpeg -i input.ts -filter_complex '[#0x2ef] setpts=PTS+1/TB [sub] ; [#0x2d0] [sub] overlay' -sn -map '#0x2dc' output.mkv
- Subtitle manipulation tools for Linux
- list of supported subtitle codecs:
- Metadades
/ Metadata
- FFmpeg Metadata
- get metadata
- ffprobe -i toto.mp4 -show_format -v quiet | grep TAG
- get creation_time:
- creation_time=$(ffprobe -i toto.mp4 -show_format -v quiet | sed -n 's/TAG:creation_time=//p')
- midentify (mplayer)
- x from 0 to ID_CLIP_INFO_N:
- ID_CLIP_INFO_NAMEx = ID_CLIP_INFO_VALUEx
- Now playing... para mplayer (bash shell to extract metadata, xosd)
- x from 0 to ID_CLIP_INFO_N:
- set metadata
- ffmpeg -i toto.mov -metadata title="Title of videoclip" ... -acodec copy -vcodec copy toto_metadata.mov
- copy metadata
- ffmpeg -i toto.mp4 -map_metadata 0 -c copy toto_out.mp4
- Bug
- Supplying FFmpeg With Metadata
- Atomic Parsley (MPEG-4)
- How to create and use Flash video metadata to add cue-points with flvtool2
- Editing video metadata
- Adobe XMP metadata (wp)
-
mminfo
QT/MOV/MP4
AVI
Matroska
MP3
MP4
MPEG TS
NUT
Realmedia
SDP
title
title Title title title
title title title s=
artist
author Artist
Author
author
author author
composer
album Album
album
year
year
track Track
track
comment Comment
comment
comment
language
language
genre Genre
genre
copyright copyright Copyright
copyright
copyright copyright
description
description
synopsis
show
episode_id
network
major_brand
minor_version
compatible_brands
creation_time
location
location-eng
- Sobreimpressió
/ Superimpose
- libav sobreimpressió / overlay
- How to make dynamic timestamp in ffpmeg, not static?
- Overlaying an Image on a Video Using FFmpeg on Ubuntu
- vhook (Video
hook documentation) (ffmpeg
and vhooks - Ubuntu forums)
- insert a text (strftime
format)
- ffmpeg -i toto1.vob -vhook '/usr/lib/vhook/imlib2.so -F /usr/share/fonts/TTF/VeraMono.ttf/14 -c black -x 0 -y 0 -t %A-%D-%T' toto.avi
- ffmpeg -i toto1.vob -vhook '/usr/lib/vhook/drawtext.so -f /usr/share/fonts/TTF/VeraMono.ttf -c #000000 -x 0 -y 0 -t Bondia' toto.avi
- insert an image
- ffmpeg -i toto1.vob -vhook '/usr/lib/vhook/imlib2.so -x 0 -y 0 -i logo.jpg' toto2.avi
- insert a text (strftime
format)
- libavfilter
- Conversió automàtica / Automatic conversion
- Enregistrament / Recording
- [FFmpeg-user]
Best
format for arbitrary start/stop stream recording?
- ffmpeg -y -isync -i INPUT -vcodec copy -acodec copy -f mpegts pipe:1 | tee -a OUTPUT1 OUTPUT2 ... OUTPUTn > /dev/null
- [FFmpeg-user]
Best
format for arbitrary start/stop stream recording?
- Streaming
(see also libav
streaming)
- Streaming guide (ffmpeg wiki)
- FFmpeg ? the swiss army knife of Internet Streaming ? part IV
- Buffers:
-
buffer_size
fifo_size
infbuf
s->buffer_size s->circular_buffer_size
transmission
- setsockopt(... SO_SNDBUF, ...)
- default: 32kB (UDP_TX_BUF_SIZE)
- parameter in command:
- -f rtp -buffer_size 212992 rtp://234.5.6.7:5100
- ...
-
-
reception
- setsockopt(... SO_RCVBUF, ...)
- default: 64kB (UDP_MAX_PKT_SIZE)
- parameter in command:
- "... -i rtp://234.1.2.3:5100?buffer_size=212992..."
- "... -i udp://234.1.2.3:5100?buffer_size=212992 ..."
- -buffer_size 212992 -i bbb.sdp
- default: 7 * 4096 * 188 bytes
- parameter in command:
- "... udp://127.0.0.1:1234?fifo_size=... "
- -fifo_size 64k -i bbb.sdp
- only for ffplay
- Pèrdua de paquets RTP / RTP packet loss
- Per a veure la mida del buffer que s'ha establert
finalment / To see the effectively set receiving or
sending buffer size (double the specified
buffer_size):
- ffmpeg/ffplay -v
debug ...
- [udp @ 0x7f4ddc00e8c0] end receive buffer size reported is 131072
- ffmpeg/ffplay -v
debug -buffer_size 65536 ...
- [udp @ 0x7f4ddc00e8c0] end receive buffer size reported is 131072
- ffmpeg/ffplay -v
debug -buffer_size 212992 ...
- [udp @ 0x7f4ddc00e8c0] end receive buffer size reported is 425984
- ffmpeg/ffplay -v
debug ...
- El valor màxim que es pot especificar a
buffer_size està limitat pel valor del paràmetre del
kernel / Maximum value than can be specified
by buffer_size option is maximum value limited by kernel
parameters:
- sudo sysctl net.core.rmem_max
- net.core.rmem_max = 212992
- Si especifiqueu un valor més gran que
net.core.rmem_max, se us mostrarà un avís / If
you specify a value greater than
net.core.rmem_max you will get a warning
message:
- sudo sysctl net.core.rmem_max
- net.core.rmem_max = 212992
- ffmpeg -v debug -buffer_size=851968 ...
- [udp @ 0x15a2280] attempted to set receive buffer to size 851968 but it only ended up set as 425984
- sudo sysctl net.core.rmem_max
- Per a poder incrementar el buffer_size més
enllà de net.core.rmem_max, haureu de modificar
el valor del paràmetre
del kernel / To increase buffer_size
beyond net.core.rmem_max you need to increase
the kernel
parameter value:
- sudo sysctl -w net.core.rmem_max=851968
- ffmpeg -v debug -buffer_size=851968 ...
- [udp @ 0x1782280] end receive buffer size reported is 1703936
- sudo sysctl net.core.rmem_max
-
- emission (use pkt_size=1316 to allocate 7 TS packets
into one IP packet: 188*7=1316 < 1500):
- ffmpeg -i toto -f mpegts udp://...:port?option1=value1&option2=value2
- ffmpeg -f video4linux2 -i /dev/video0 -an -s 352x288 -vcodec mpeg2video -f mpegts udp://172.16.2.146:1234?pkt_size=1316
- ffmpeg -f video4linux2 -i /dev/video0 -s 352x288 -f mpegts -vcodec libx264 -vpre ultrafast -vpre baseline udp://172.16.2.146:1234?ttl=10&buffer_size=65536&pkt_size=1316
- RTP
- Jearquia
de streaming / Streaming hierarchy
- SDP
- Fragmentation and
packet size
ffmpeg -f rtp rtp://...?pkt_size=1472-
headers (bytes)
RTP (pkt_size)
ethernet IP UDP header
payload
14
20
8
12
1460
1472 (default pkt_size)
1500 (default MTU for ethernet)
1514
- when streaming to Janus (WebRTC) streaming plugin, the value of pkt_size must be lower than 1472 (e.g. 1400). Otherwise the addition of headers for SRTP will overpass the MTU limit, and NACKs will appear.
- No coding (from file)
- Using SDP with increased buffer_size
- transmission (if not specified,
buffer_size is 32kB (UDP_TX_BUF_SIZE))
- ffmpeg -v debug -protocol_whitelist rtp,udp,file,http,https,tcp,tls,rtmp -re -i bbb_stereo.mp4 -an -c:v copy -f rtp -buffer_size 128k rtp://234.5.6.7:5100 -vn -c:a copy -f rtp -buffer_size 256k rtp://234.5.6.7:5102 -sdp_file bbb_stereo.sdp
- with specified RTP packet size (32000
bytes) (not recommended, as network
errors imply worse consequences):
- ffmpeg -v debug -protocol_whitelist rtp,udp,file,http,https,tcp,tls,rtmp -re -i bbb_stereo.mp4 -an -c:v copy -f rtp -buffer_size 128k rtp://234.5.6.7:5100?pkt_size=32000 -vn -c:a copy -f rtp -buffer_size 256k rtp://234.5.6.7:5102?pkt_size=32000 -sdp_file /mnt/nfs/sdp/bbb.sdp
- reception (if not specified, buffer_size
is 64kB (UDP_MAX_PKT_SIZE))
- ffplay -v debug -protocol_whitelist rtp,udp,file,http,https,tcp,tls,rtmp -buffer_size 256k -i bbb.sdp
- transmission (if not specified,
buffer_size is 32kB (UDP_TX_BUF_SIZE))
- Using SDP with increased buffer_size
- Coding
- H.264
- ffmpeg -re -f lavfi -i testsrc=size=320x240:r=15 -c:v libx264 -g 15 -profile:v baseline -level 31 -keyint_min 15 -sc_threshold 0 -b:v 1M -pix_fmt yuv420p -f rtp rtp://127.0.0.1:8004 >/tmp/barres_h264.sdp
- H.264 / AAC
- ffmpeg -re -f lavfi -i testsrc=size=320x240:r=15 -f lavfi -i sine=f=440:b=1 -c:v libx264 -profile:v baseline -b:v 1M -pix_fmt yuv420p -an -f rtp rtp://224.1.2.3:8004 -c:a aac -b:a 16k -vn -f rtp rtp://224.1.2.3:8002 >/tmp/barres_h264_aac.sdp
- VP8 / Opus (e.g. for Janus
streaming plugin with default example
janus.plugins.streaming.cfg.sample.in)
- ffmpeg -re -f lavfi -i testsrc=size=320x240:r=15 -f lavfi -i sine=f=440:b=1 -c:v vp8 -b:v 1M -pix_fmt yuv420p -an -f rtp -payload_type 100 rtp://224.1.2.3:5004 -c:a opus -b:a 16k -vn -f rtp -payload_type 111 rtp://224.1.2.3:5002 -sdp_file /tmp/barres_vp8_opus.sdp
- ffmpeg -re -f lavfi -i testsrc=size=320x240:r=15 -f lavfi -i sine=f=440:b=1 -c:v vp8 -b:v 1M -pix_fmt yuv420p -an -f rtp rtp://224.1.2.3:5004 -c:a opus -b:a 16k -vn -f rtp rtp://224.1.2.3:5002 >/tmp/barres_vp8_opus.sdp
- H.264
- Specify payload
type (pt):
ffmpeg ... -f rtp -payload_type 100 rtp://...
- Specify CNAME, to be included in RTCP SR "Source
Description" (SDES)
packets:
... -f rtp -cname user123456789@host-87654321 rtp://...- Source Description in RTCP packets is needed by gstreamer as an RTP client, to honour RTCP incormation
- Specify title, to be included in SDP as
s=<title> - Problemes / Problems
- Pèrdua
de paquets en recepció / Packet loss in
reception
- Info
- GStreamer RTP
- FFMPEG RTSP streaming over UDP - Image Smearing
- Live555 FAQ
- Deployment guide (liveMediaStreamer)
- max delay reached. need to consume packet
RTP: missed xx packets - Solució / Solution
- tune
buffers: buffer_size, fifo_size
- e.g.: if receiving buffer_size is
65536 (ffmpeg default, or specified
with buffer_size 65536)
- [udp @ 0x7f548c00e8c0] end receive buffer size reported is 131072
- NALs larger than 65536*2=131072 will be truncated, and rtp packets from this point will be dropped, and kernel will mark them as dropped (it can be seen by running: dropwatch -l kas)
- maximum value for buffer_size
depends on Linux
kernel network parameter:
- check that net.core.rmem_max is, at least, 212992 (default)
- ffmpeg -buffer_size=212992
- this will open an udp buffer of 212992*2=425984 and NALs could have up to 425984 bytes (416kB)
- e.g.: if receiving buffer_size is
65536 (ffmpeg default, or specified
with buffer_size 65536)
- tune
buffers: buffer_size, fifo_size
- Info
- En recepció / In reception
- [h264 @ 0x1795060] co located POCs
unavailable
- Solució / Solution
- ...
- Solució / Solution
- [h264 @ 0x1795060] co located POCs
unavailable
- En transmuxació / Transmuxing
- No H.264
NAL units of type
7 (SPS), 8 (PPS)
when transmuxing from RTMP
(at RTMP: NAL units of type 7,8 are only
available at the beginning of the
connection, but they are needed by RTP
client)
- Solució / solution
ffmpeg -protocol_whitelist rtp,udp,file,http,https,tcp,tls,rtmp -re -i rtmp://... -an -bsf:v h264_mp4toannexb -c:v copy -f rtp rtp://...
- Solució / solution
- Too many packets buffered for output
stream 0:1
- Solució / Solution
- Set a high value for -max_muxing_queue_size
- [bug][regression] Too many packets buffered for output stream
- Solució / Solution
- No H.264
NAL units of type
7 (SPS), 8 (PPS)
when transmuxing from RTMP
(at RTMP: NAL units of type 7,8 are only
available at the beginning of the
connection, but they are needed by RTP
client)
- Pèrdua
de paquets en recepció / Packet loss in
reception
- Jearquia
de streaming / Streaming hierarchy
- RTMP
- rtmp
- ffserver accepts RTMP input - can it output to RTMP?
- from inverted camera to an rtmp server:
ffmpeg -f video4linux2 -i /dev/video1 -s 640x480 -f flv -c:v libx264 -profile:v baseline -pix_fmt yuv420p -an -vf "vflip,hflip" -r 30 rtmp://...
- from file to an rtmp server:
ffmpeg -re -i sintel-1024-stereo.mp4 -c:v copy -c:a aac -strict -2 -f
flv rtmp://...
- from an rtmp server:
ffmpeg -i rtmp://...- Problemes /
Problems
- if ffmpeg is not receiving data from
the rtmp server (because the process
that feeds it has not cleanly
terminated), it cannot be terminated
(kill -TERM) (see nginx
rtmp module)
- workarounds:
- kill -TERM again
- send some content to the rtmp server
- if ffmpeg is not receiving data from
the rtmp server (because the process
that feeds it has not cleanly
terminated), it cannot be terminated
(kill -TERM) (see nginx
rtmp module)
- Problemes /
Problems
- low-latency
- Latency (Streaming guide)
- Format options
- libav x264 settings for zero-latency encoding (arash@work)
- compile the latest ffmpeg and libx264
- export LD_LIBRARY_PATH=/usr/local/lib; /usr/local/bin/ffmpeg -an -f video4linux2 -s 640x480 -r 15 -i /dev/video0 -vcodec libx264 -preset ultrafast -tune zerolatency -x264opts "vbv-maxrate=2000:vbv-bufsize=200:slice-max-size=1500" -f h264 - | /usr/local/bin/ffplay -i -
- x264: the best low-latency video streaming platform in the world
- x264 encoder latency
- Streaming x264 zerolatency
- x264 zerolatency - How to do it?
- Using WebM/VP8 for low latency streaming
- reception:
- allow reuse multicast packets (UDP sockets) in reception (see also LIMBOS / mcreflect, Darwin reflector) (if not, only one application per computer will be able to receive the stream):
- ffmpeg ... -i udp://...?reuse=1
- vlc udp://@:1234
- UDP
- ffmpeg -i udp://...?fifo_size=... (similar to -infbuf for ffplay/avplay?)
- ffmpeg live stream transcoding
- Tutorial:
HD-PVR
to Twitch on Mac OS with ffmpeg
- ffmpeg -i 'udp://127.0.0.1:1234?fifo_size=1000000 &buffer_size=10000000 &overrun_nonfatal=1? ...
- HLS
- hls muxer
- hls demuxer (ffmpeg documentation)
- Segmentació / Segmentation
- #5419 HLS EXT-X-DISCONTINUITY tag is not supported
- Exemple / Example
ffmpeg -i toto.mp4 -c copy -hls_time 2 -hls_list_size 0 -start_number 0 out.m3u8
- MPEG
DASH
- compile
with options:
--enable-demuxer=dash --enable-libxml2
- Reproducció / Play
ffplay toto.mpd
- Selecció de qualitat / Quality selection
ffprobe toto.mpdffplay -map ... -map ... toto.mpd
- WebM DASH
- compile
with options:
- Problemes / Problems
- RTMP problems
- RTP problems
- Tee problems
- Timestamps problems
- HLS problems
- empty m3u8 after termination
- from v4.4
- Empty m3u8 after a Ctrl+C (#9234)
- EXT-X-PROGRAM-DATE-TIME:1970-01-01...
[https @ 0x1adbbe0] Stream ends prematurely at xxxx, should be yyyy- Solució / Solution
ffmpeg -reconnect 1 ...
- Solució / Solution
Past duration ... too large- #4700 Past duration 0.999992
- #4401
Past duration
- Only supress warnings:
- -loglevel error
- Possible solutions
- -filter:v fps=(your framerate)
- -vsync 1 (cfr)
- -vsync passthrough -i ...
- -vsync passthrough -frame_drop_threshold 4 -i ...
- Only supress warnings:
- Strange behaviours:
- frame-rate
conversion:
- ffmpeg -i toto.m2v -r 12.5 toto.h264
- ffmpeg -i toto.h264: ...12.50 tb(r)
- vlc plays toto.h264 2x too fast; mplayer plays it correctly
- ffmpeg -i toto.h264 -vcodec copy -r 12.5 toto.mp4
- mp4info toto.mp4: ...@25.00 fps
- vlc and mplayer play toto.h264 2x too fast;
- ffmpeg -r 12.5
-i toto.h264 -vcodec copy toto_12.5.mp4
- mp4info toto_12.5.mp4: ...@12.500 fps
- vlc and mplayer play toto.mp4 correctly
- ffmpeg -i toto.m2v -r 12.5 toto_direct.mp4
- mp4info toto_direct.mp4: ...@12.580 fps
- vlc and mplayer play toto.mp4 correctly
- ffmpeg -i toto.m2v -r 12.5 toto.h264
- frame-rate
conversion:
- Info
Subtítols / Subtitles
- Conversió de subtítols / Subtitle conversion
- TV subtitles
- Validadors
- Estàndards / Standards
- DVB subtitles
- SMPTE Timed Text
- EBU TT
- W3C TTML
- CEA-708 Digital Television (DTV) Closed Captioning
- 3GPP Timed Text
- WebVTT
- SRT
- verify
Còdecs vídeo / Video codecs
- Codecs (Videolan Wiki)
- Video codecs (Wikipedia)
- Video codecs (MultimediaWiki)
- Royalty-free codecs (Open Video Alliance)
- List of open source codecs (wp)
- Video encoding (Doom9)
- Llicències / Licenses
- MPEG-LA
- Via Licensing
- Licensing issues with H.264 video
- MPEG LA: H.264 streaming will be free forever
- Linux
- The role of legal codecs in Linux installs (pdf) (Fluendo)
- FAQ (Fluendo)
- Videolan
FAQ (pdf) (Videolan)
- 3.4. What about personal/commercial usage?
- Comercial / Commercial
- HTML5
video
- HTML5 & Standards
- The State of HTML5 Video (JW Player)
- HTML5 video and codecs
- HTML5, ¿Con que códec nos quedamos H.264 o VP8?
- YouTube HTML5
- YouTube y Vimeo soportan HTML5 ¿comienza el fin de Flash?
- Decoding the HTML 5 video codec debate
- Google to Open-source VP8 for HTML5 Video
- Video for everybody
- HTML5 video (w3schools)
- "Towards Video on the Web with HTML5" (NEM Summit 2010)
- HTML5Video (demos)
- HTML5 video tutorial (HTML5Rocks)
- HTML5 Video Events and API (demo: playbackRate, ...) (requires Chrome/Chromium)
- Video on the web (W3c presentation)
- HTML encoding secrets (video)
- Video on the Web
-
container
video
audio
Firefox
Opera
Chromium
Chrome
Safari
iPhone
Android
WebM
VP8
Vorbis
>=4
>=10.60
x
>=6
Ogg
Theora
Vorbis
x
x
x
x
-
-
-
MP4
H.264
AAC
-
-
-
x
x
x
x
- Videollamada en WebKit directamente desde HTML5
- Reproductors / Players
- Edició / Edition
- Streaming
- WebM streaming
- Why Flash is better than HTML5 for your live webcast (browsers table)
- Streaming via RTSP or RTP in HTML5
- Real-time video streaming with HTML5 (pdf) (Master's Thesis | Lund University 2014)
- Demos, tutorials
- Moving the Timeline Position (Tizen)
- Jumping
to
time offsets in HTML5 video
- currentTime
- timeupdate
- Media Fragments
- Video Timeline using HTML5 and javascript
- Subtítols / Subtitles
- WebVTT (Web Video Text Tracks)
- Info
- WebVTT and Video Subtitles
- Video Subtitling and WebVTT (HTML5 Doctor)
- Subtitles, Captions, WebVTT, HLS, and those magic flags (Mux)
- Recent developments around WebVTT (2011)
- Getting Started With the Track Element
- JavaScript: HTML5 Video with SRT Subtitles
- Understanding WebVTT file format (draft)
- WebVTT Presentation
- Demos
- HTML Video books (code) (Silvia Pfeiffer)
- Positioning WebVTT In Firefox
- Using JSON with the HTML5 video track element
-
text track kind
track element
subtitles
captions
descriptions
chapters
metadata
- Specification
- Structure
WEBVTT
[STYLE ...]
[REGION ...]
[NOTE ...]
[cue_id]
[hh:]MM:ss.mmm --> [hh:]MM:ss.mmm [cue_settings]
[tag_begin]text line[tag_end]
[tag_begin]text line[tag_end]
...
[NOTE ...]
[cue_id]
[hh:]MM:ss.mmm --> [hh:]MM:ss.mmm [cue-settings]
[tag_begin]text line[tag_end]
[tag_begin]text line[tag_end]
...- cue-settings
- tags (cue
components)
-
tag
description
example
b
bold
<b>text</b>
c
class
<c.my_class_name>
i
italic
lang
language
ruby
ruby
rt
ruby text
u
underline
v
voice
-
- STYLE
::cue(v[voice="Esme"]) { color: cyan }
::cue(v[voice="Mary"]) { color: lime }
::cue(i) { font-style: italic }
::cue(.loud) { font-size: 2em }
- REGION
- Conversió / Conversion
- from srt to webvtt:
ffmpeg -i toto.srt -f webvtt toto.webvtt- Players
- HTML5 video
<video width="640" height="480" controls>
<source src="video.mp4" type="video/mp4" />
<source src="video.webm" type="video/webm" />
<track src="subtitles.vtt" kind="subtitles" srclang="en" label="English" />
<!-- fallback for rubbish browsers -->
</video>
- Navegadors / Browsers
- WebVTT - Web Video Text Tracks (CanIuse)
- Tracks & WebVTT (JWPlayer)
- Firefox
- Adding
captions
and subtitles to HTML5 video (MDN)
- close captions menu in controls
- about:config
media.webvtt.enabledmedia.webvtt.regions.enabled
- Bug 981280 - WebVTT / HTML5 track support too buggy to switch it on
- Adding
captions
and subtitles to HTML5 video (MDN)
- Problems
Access to Text track at 'file:///...vtt' from origin 'null' has been blocked by CORS policy: Invalid response. Origin 'null' is therefore not allowed access.- Solució / Solution
- use Firefox
- controls?
- use an http server
google-chrome --disable-web-security --user-data-dir=$HOME/.config/google-chrome/Default
- use Firefox
- Solució / Solution
- hls.js
- HTML5 video
- Info
- WebVTT (Web Video Text Tracks)
- Sincronització / Synchronisation
- Standards
- Timing object (W3C)
- Implementacions / Implementations
- Shared Motion (JavaScript) (Motion Corporation)
- Carnival (yt)
- Standards
- Position and timeline
- HTML5 audio
| file ext | file (UNIX) | MPEG-4 RA | FOUR CC | license |
used
by |
||||
|---|---|---|---|---|---|---|---|---|---|
| handler | codec (1) | ||||||||
| MPEG | MPEG-1 | part 2 |
|||||||
| MPEG-2 | part 2 |
.m2v | MPEG sequence, v2, MP@ML progressive Y'CbCr 4:2:0 video,
CCIR/ITU PAL 625, 25 fps |
MP2V | |||||
| MPEG-4 | part 2: [advanced] simple profile |
.m4v | MPEG sequence, v4, video, [advanced] simple | vide | mp4v | MP4V | |||
| part 10: AVC/H.264 | .264 | JVT NAL sequence, H.264 video | avc1 | AVC1 | FLV, F4V, ... |
||||
| avc2 |
|||||||||
| avc3 (amdt2)(*) |
|||||||||
| svc1 |
|||||||||
| Apple |
ProRes (wp) |
||||||||
| Avid |
DNxHD (wp) | ||||||||
| Microsoft | WMV | ||||||||
| VC-1 | vc-1 | ||||||||
| Real Networks | RealVideo | ||||||||
| Xiph.org |
Theora (based on VP3) |
||||||||
| BBC |
Dirac (VC-2) | drac | - | ||||||
| Streambox |
ACT-L3 |
||||||||
| Adobe |
Flash Video |
FLV1 |
|||||||
| On2 (Google) |
VP3 |
Theora | |||||||
| VP6 (wp) |
|||||||||
| with alpha |
VP6A |
FLV |
|||||||
| VP8 |
|||||||||
| Sorenson |
Sorenson Video / SVQ |
QuickTime |
|||||||
| Sorenson Spark / H.263 |
FLV | ||||||||
- As shown by:
- vlc: Tools / Codec information
- mplayer: "VIDEO: [xxxx]"
- (ffmpeg shows "h264" instead of avc1)
- H.264
- On2 (Google) (wp)
- Dirac
(BBC
Research) (wp)
urpmi vlc-plugin-schroedinger dirac-utils- Dirac
Pro ?Light Compression? (pdf) (BBC) (EBU Technical)
- Dirac-Pro users: ORF (Austrian national broadcaster), Westminster Live, Asia Television ? Hong Kong, Television Broadcasts Limited (TVB) ?Hong Kong,Rainbow Media ? New York, Cambridge University & Essex University, BBC ? Beijing Olympics, OBs, Pacific Quay
- Evaluation of Dirac and Theora
- Performance Comparison of Emerging Dirac Video Codec with H.264/AV (IEEE): quality factor.
- COMPARISON OF OPEN AND FREE VIDEO COMPRESSION SYSTEMS A Performance Evaluation Till Halbach
- Theora (wp)
- Cortado (Java applet decoder) (Fluendo)
- YouTube / Ogg/Theora comparison
- ACT-L3 (Streambox)
- VPX: VP8, VP9
- WebM Project
- Users (supported players and browsers)
- Specs
- RTP payload
- libvpx
compilation
- Dependencies
- Mageia
urpmi yasm
- Mageia
- get source (one of the following):
git clone https://chromium.googlesource.com/webm/libvpxgit clone https://github.com/webmproject/libvpxmkdir libvpx
cd libvpx
version=v1.9.0
#version=v1.8.2
#wget https://chromium.googlesource.com/webm/libvpx/+archive/${version}.tar.gz
# https://github.com/webmproject/libvpx/archive/v1.9.0/libvpx-1.9.0.tar.gz
wget https://github.com/webmproject/libvpx/archive/v${version}/libvpx-${version}.tar.gz
tar xvzf ${version}.tar.gz
cd libvpx
./configuremakemake install- (compile
ffmpeg with
--enable-libvpx) - Usage:
ffmpeg ... -c:v libvpx_vp8 ...ffmpeg ... -c:v libvpx_vp9 ...
- Problemes / Problems
test/sad_test.cc:543:3: note: in expansion of macro ?DECLARE_ALIGNED?
DECLARE_ALIGNED(kDataAlignment, uint32_t, exp_sad[4]);- Issue 1690: Can't build using GCC 4.8.5/centOS
- Issue 1690 in webm: Can't build using GCC 4.8.5/centOS
- Build FFmpeg almost fully automatically on Linux (compatible with CentOS, Ubuntu, etc.) (original in Japanese)
- Solució / Solution
- ...
- Dependencies
- HTML5 video
- VP8 Data Format and Decoding Guide (IETF RFC draft-bankoski-vp8-bitstream)
- WebM VP8 vs. MPEG-4/H264 (*)
- Technical Overview of VP8, an Open Source Video Codec for the Web (AVCC / ICME 2011)
- The first in-depth technical analysis of VP8 (Diary Of An x264 Developer)
- H.264/AVC and VP8 performance analysis for consumer videoconference applications (STM) (EBU Broadthinking webinar)
- On2 VP8
- Rate distortion performance of VP8 (WebP and WebM) when compared to standard image and video compression techniques (MMSPG EPFL)
- NewTeeVee
- WebM Project
- Multiview
- Eye Vision (CMU / Super Bowl)
- iview (BBC)
- Multi view demo
- Apple 4K
- Low bit-rate
- Question about HDTV: several resolutions at 1Mbps
Còdecs àudio / Audio Codecs
- Audio codecs (Wikipedia)
- Digital Audio Formats (The MP3 and Internet Audio Handbook)
- The Computer Music Guide to Digital Audio Formats
- Examples of audio coding (AC-2, AC-3, PAC, ATRAC)
- Leopold's Comparison Between Audio Formats
- Samples per frame:
samples/frame
sample format
info
standard
flag
MPEG-1,2
384
576
1152 s16p
- 3 groups * 12 samples * 32 subbands
- MP3 (Hydrogenaudio knowledgebase)
MPEG-4, AAC
1024
fltp
ISO/IEC 14496-3 subpart 4. frameLengthFlag=0 960
frameLengthFlag=1 AC3
1536
- 6 groups * 256 samples
- Dolby Digital Audio CD explained
| MPEG | 1 | layer 3 | MP3 | |||||||||||||
| 2 | part 7 |
|
||||||||||||||
| 4 | part 3 | |||||||||||||||
| 4 |
|
Streaming
Còdecs d'imatges / Image codecs
- WebP (Google, based on VP8/webM)
- Cru / Raw
- Visualitzador / Viewer
- ImageMagick
http://www.francescpinyol.cat/formats_av.html
Primera versió / First version: 3.IV.2006
Darrera modificació: 11 de gener de 2023 / Last update: 11th
January 2023
Cap a casa / Back home.