GStreamer
Índex
- General
- Instal·lació
/
Installation
- Ús / Usage
- Edició / Edit
- Acceleració / Acceleration
- Raspberry Pi
- CLI tools
-
- gst-validate
- ges-launch
- gst-transcoder
- gst-discoverer
- gst-inspect
- gst-launch
- gst-validate
- Desenvolupament / Development
-
- Documentació / Documentation
- Llenguatges / Languages
- gst
- Resum / Summary
- Application Development Manual
- About GStreamer
- Building an application
- Advanced GStreamer concepts
- Position tracking and seeking
- Metadata
- Interfaces
- Clocks and synchronization in GStreamer
- Buffering
- Dynamic controllable parameters
- Threads
- Autoplugging
- Pipeline manipulation
- Higher-level interfaces for GStreamer applications
- Appendices
- Tutorials
- Encoding
- Encoding target
- Encoding profile
- Encoding preset
- gst-editing-services (GES)
- gst-transcoding
- Documentació / Documentation
-
General
- GStreamer (.com) (wp)
- Releases
- Latest stable release: 1.14.4
- Latest development release: 1.15.1
- Plug-ins
- GStreamer SDK documentation
- Development
-
- Developer information
-
- modules
- stable version
- cgit.freedesktop.org/gstreamer/
- Gstreamer cheat sheet
- Introducing GStreamer (CInLUG)
Instal·lació / Installation
- From packages
-
- Mageia
-
urpmi ... gstreamer1.0-plugins-bad gstreamer1.0-plugins-ugly gstreamer1.0-x264
- Compilació des de codi font / Compilation from source
-
- Mòduls i dependències / Modules and dependencies
-
-
module desc
dependencies
Mageia CentOS Ubuntu
/data/doc/gstreamer/head
urpmi ...
yum install ...
apt-get install ...
gstreamer
autoconf gettext-devel libtool bison flex gtk-doc yasmautoconf gettext-devel libtool bison flex gtk-doc yasm glib2-devel gcc-c++ freetype freetype-devel
autoconf bison flex...
gst-plugins-base
lib64opus-devel libvorbis-devel libogg-devel libtheora-devel libxv-devel
opus-devel libvorbis-devel libogg-devel libtheora-devel libxv-devel pango-devel wayland-devel
libopus-dev libvorbis-dev libogg-dev libtheora-dev libxv-dev libpango1.0-dev
gst-plugins-good
libvpx-devellibsoup-devellibvpx-devel pulseaudio-libs-devel libsoup-devel
libvpx-devgst-plugins-bad
librtmp-devellibrtmp-devellibrtmp-devgst-plugins-ugly
libx264-devellibx264-devellibx264-devgst-python
python-gobject3-devel
python-devel pygobject3-devel
gst-libav
gstreamer-editing-services
libxml2-devellibxml2-devel
- Mageia
-
urpmiautoconf gettext-devel libtool bison flex gtk-doc yasm
- For plugins-base:
-
urpmi lib64opus-devel lib64vorbis-devel lib64ogg-devel lib64theora-devel lib64xv-devel libsoup-devel
- Raspbian
-
- ...
- CentOS
-
- automake >=1.14
(CentOS 7 provides version 1.13)
yum install -y autoconf gettext-devel libtool bison flex gtk-doc yasm glib2-devel gcc-c++ freetype freetype-devel
- plugins-base
-
yum install opus-devel libvorbis-devel libogg-devel libtheora-devel libxv-devel pango-devel
- plugins-good
-
yum install libvpx-devel
- plugins-bad
-
yum -y install http://li.nux.ro/download/nux/dextop/el7/x86_64/nux-dextop-release-0-5.el7.nux.noarch.rpm
yum -y install librtmp-devel
- plugins-ugly
-
yum install libx264-devel
- automake >=1.14
(CentOS 7 provides version 1.13)
-
- Modules
-
module
/usr/local/lib/
/usr/local/lib/gstreamer-1.0/ /usr/local/lib/pkgconfig/ /usr/local/lib/girepository-1.0/
/usr/local/share/gir-1.0/
gstreamer
libgstreamer-1.0.so
libgstbase-1.0.so
libgstcheck-1.0.so
libgstcontroller-1.0.so
libgstnet-1.0.so
libgstcoreelements.so
libgstcoretracers.so
gstreamer-1.0.pc
gstreamer-base-1.0.pc
gstreamer-check-1.0.pc
gstreamer-controller-1.0.pc
gstreamer-net-1.0.pcGst*1.0.typelib
Gst-1.0.gir
GstBase-1.0.gir
GstController-1.0.gir
GstNet-1.0.gir
GstCheck-1.0.gir
...
gstreamer-editing-services
libges-1.0.so
gst-editing-services-1.0.pc GES-1.0.typelib
GES-1.0.gir
- From git
- From tar files
-
- https://gstreamer.freedesktop.org/src/
- if you want to be able to access GStreamer by using PyGObject (applications made in Python):
-
- Dependencies
-
- Mageia
-
urpmi lib64girepository-devel
- CentOS
-
sudo yum install gobject-introspection-devel
- Check config.log
-
HAVE_INTROSPECTION_TRUE=''
INTROSPECTION_COMPILER='/usr/bin/g-ir-compiler'
...
- Check that these files exist after compilation:
-
/usr/local/lib/girepository-1.0/Gst*.typelib
- In order to access Gst from your applications,
you will need to set environment
variable (to avoid error:
ValueError: Namespace Gst not available): -
export GI_TYPELIB_PATH=/usr/local/lib/girepository-1.0
gstreamer_install.sh1.15.1- gstreamer_install.sh
-
#!/bin/bash -e
EXPECTED_ARGS=2
if (( $# != $EXPECTED_ARGS ))
then
cat <<EOF
Usage: `basename $0` action version
Examples:
- `basename $0` get 1.15.1
- `basename $0` install 1.15.1
- `basename $0` uninstall 1.15.1
EOF
exit 1
fi
# parameters
action=$1
version=$2
mkdir -p gst-${version}
cd gst-${version}
# dependencies:
# automake 1.14
# gstreamer: autoconf gettext-devel libtool bison flex gtk-doc yasm glib2-devel gcc-c++ freetype freetype-devel
# plugins-base: libogg-devel libtheora-devel libvorbis-devel opus-devel wayland-devel
# plugins-good: libvpx-devel pulseaudio-libs-devel libsoup-devel
# plugins-bad: http://li.nux.ro/download/nux/dextop/el7/x86_64/nux-dextop-release-0-5.el7.nux.noarch.rpm librtmp-devel
# plugins-ugly: libx264-devel
# gst-python: python-devel pygobject3-devel
# gstreaming-editing-services: libxml2-devel
# update ldconfig
sudo sudo sh -c 'echo "/usr/local/lib" > /etc/ld.so.conf.d/local.conf'
sudo ldconfig
modules="gstreamer gst-plugins-base gst-plugins-good gst-plugins-bad gst-plugins-ugly gst-libav gst-python gstreamer-editing-services"
for module in $modules
do
src_name="${module}-${version}"
echo "========================================= $src_name========================================="
case $action in
"get")
curl -s -L -O https://gstreamer.freedesktop.org/src/${module}/${tar_filename}
;;
"install")
tar_filename="${src_name}.tar.xz"
tar xJf ${tar_filename}
export XDG_DATA_DIRS="/usr/local/share/:/usr/share/"
cd ${src_name}
./autogen.sh PKG_CONFIG_PATH=/usr/local/lib/pkgconfig/ --disable-gtk-doc
make
sudo make install
sudo ldconfig
cd ..
;;
"uninstall")
cd ${src_name}
sudo make uninstall
cd ..
;;
esac
done
exit 0
- gstreamer will be installed in:
-
/usr/local/lib/gstreamer-1.0/
- https://gstreamer.freedesktop.org/src/
- Problemes / Problems
-
- 1.16.0
- install wayland-devel
- autotools gtk-doc scanner build failure: undefined reference to `gst_gl_display_wayland_get_type'
- Solució / Solution
./autogen.sh PKG_CONFIG_PATH=/usr/local/lib/pkgconfig/ --disable-gtk-doc.../tmp-introspectAx4R5G/.libs/lt-GstMpegts-1.0: error while loading shared libraries: libgstvideo-1.0.so.0: cannot open shared object file: No such file or directory
- Solution
sudo ldconfig
(gst-plugin-scanner:21672): GStreamer-WARNING **: 15:56:30.108: Failed to load plugin '/usr/local/lib/gstreamer-1.0/libgstpango.so': /lib64/libcairo.so.2: undefined symbol: FT_Get_Var_Design_Coordinatesgst-inspect-1.0 timeoverlay
- Solució / Solution
- update freetype to 2.8
- Opus audio encoder not found
-
gst-plugins-bad/ext/opus/opusenc.cwas present until version 1.6; then it was moved to plugins-base- Solució / Solution
-
- install system-wide Opus devel libraries and reconfigure and make plugins-base
Ús / Usage
- Edició / Edit
- Info
- splitting
/ segmenting video stream with gstreamer (gnonlin)
- "Cutting" Video via GStreamer (GES ges-launch-1.0)
- How
do I use gstreamer to make an audio clip from a segment of a longer
source? (gnonlin)
- Eines / Tools
- ges-launch
- gst-launch
- Desenvolupament / Development
- ...
- Acceleració / Acceleration
- Raspberry Pi
-
- Hardware Video Encoding progess with the Raspberry Pi
- GStreamer
1.2 on the Raspberry Pi
- Using Gstreamer with OpenMax in a Raspberry Pi
- Hardware-accelerated video playback on the Raspberry Pi
- Streaming
with nginx-rtmp-module: GStreamer and Raspberry Pi
- Using GStreamer as transcoder for Dreambox or TVheadend
- Usage of hardware-accelerated encoding/decoding:
-
- decoding:
-
- omxmpeg2videodec (license needed)
- omxh264dec
- encoding:
-
- omxh264enc
-
gst-inspect-1.0 omxh264enc
- Using GStreamer as transcoder for Dreambox or TVheadend
-
- Bug 720031 -
omxh264enc: key frame interval missing
- solution
to missing SPS/PPS
-
- http://pastebin.com/b5avcxpm (already in the latest git version)
- Bug 720031 -
omxh264enc: key frame interval missing
- decoding:
- Tools
(CLI)
- How do I use the GStreamer command line interface ?
- Command line tools
- gst-validate
- ges-launch
-
- GES in Python
- Help
ges-launch-1.0 -h
- ges-videocrop-effect.sh
- Sintaxi / Syntax
-
project-related options
--load[-l]=-s --save=-p --sample-path-r --sample-path-recurse
rendering options
-o --outputuri=<uri>()
-f --format=<profile>(specified serialized encoding-profile; if not specified: application/ogg:video/x-theora:audio/x-vorbis)
-e --encoding-profile=<profile-name>(from a preset file)
playback options
-v --videosink=-a --audiosink=-m --mute
helpful options
--inspect-action-type=--list-transitions
generic options
--disable-mixing-r --repeat=--set-scenario
+clip
<path|uri>inpoint[i]=duration[d]=start[s]=layer[l]=set-alphaposxposywidthheightvolumemute
+test-clip
smptecircular
...
+effect
<bin-description>agingtvvideocrop...
set-
+title
<text>
- Exemples / Examples
- play a clip from second 4.0 to second 6.0:
ges-launch-1.0 +clip bbb_720p.mp4 i=4.0 d=2.0- play a clip from second 4.0 to second 6.0 and then another clip from the beginning:
ges-launch-1.0 +clip bbb_720p.mp4 i=4.0 d=2.0 +clip sintel_720p.mp4- play a clip with a logo during 10 seconds on top right:
ges-launch-1.0 +clip bbb_720p.mp4 +clip logo.jpeg s=0 d=10 set-alpha 0.8 set-width 200 set-height 100 set-posx 1000 set-posy 20
- save project to play a clip from second 4.0 to second 6.0:
ges-launch-1.0 +clip bbb_720p.mp4 i=4.0 d=2.0 --save bbb.xges- play according to project (can also be generated by pitivi):
ges-launch-1.0 --loadbbb.xges- launch pitivi with this project:
pitivibbb.xges- render to Ogg - Theora - Vorbis (default encoding profile):
ges-launch-1.0 +clip bbb_720p.mp4 i=4.0 d=2.0 -o bbb.oggges-launch-1.0 +clip bbb_720p.mp4 i=4.0 d=2.0 -o bbb.ogg -f "video/ogg:video/x-theora:audio/x-vorbis"
- render to WebM - VP8 - Vorbis:
ges-launch-1.0 +clip bbb_720p.mp4 i=4.0 d=2.0 -o bbb.webm -f "video/webm:video/x-vp8:audio/x-vorbis"- Problemes / Problems
ERROR from element qtdemux1: Internal data stream error.- Solució / Solution:
urpmi gstreamer1.0-vp8
- render to MP4 - H.264 - MP3:
ges-launch-1.0 +clip bbb_720p.mp4 i=4.0 d=2.0 -o bbb.mp4 -f "video/quicktime,variant=iso:video/x-h264:audio/mpeg,mpegversion=1,layer=3"
- render to MP4 - H.264 - AAC encoding profile (Mageia: gstreamer1.0-plugins-bad...tainted):
ges-launch-1.0 +clip bbb_720p.mp4 i=4.0 d=2.0 -o bbb.mp4 -f "video/quicktime,variant=iso:video/x-h264:audio/mpeg,mpegversion=4"- Problemes / Problems
ERROR from element qtdemux1: Internal data stream error.- Solució / Solution
urpmigstreamer1.0-x264- specify output video dimensions (clips will not be rescaled)
ges-launch-1.0 ...-o bbb.mp4-f "video/quicktime,variant=iso:video/x-h264,width=1920,height=1080:audio/mpeg,mpegversion=1,layer=3"
- ...
- GES in Python
- gst-transcoder
gst-transcoder-1.0 [OPTION?] <source uri> <destination uri> <encoding target name>[/<encoding profile name>]
- Exemples / Examples
- Create a target file called device/mp4target.gep
gst-transcoder-1.0 <input_file> <output_file>.mp4 mp4target/mp4
gst-transcoder-1.0 input.mp4 output.mkv matroska- Problemes / Problems
WARN: ... no such element factory "uritranscodebin"!
- gst-discoverer
- gst-inspect
-
- list of all plug-ins
-
gst-inspect-1.0
- available properties for a
specified plugin
-
gst-inspect-1.0 videoconvert...
- ...
- gst-launch (wp)
-
gst-launch-1.0 ... ! ... ! ...
gst-launch-1.0 ... ! ... !...demux name=mydemux...mux name=mymux! ... ! ... mydemux. ! ... ! mymux. mydemux. ! ... ! mymux.
-
input + demuxmux + outputaudiovideo
- Options
-
-e: end of stream on shutdown-f, --no_fault: ...
--help: ...
-q, --quiet: ...
-m, --messages: ...
-o FILE, --output=FILE: ...
-t, --tags: ...
-T, --trace: ...
-v: verbose : ...--gst-debug-level=2
- Verbose messages (
-v, -q): sent to stdout (1)
/<element>:<name>/<element>:<name>.<subelement>:<name>: <property>=<value>, <property>=<value> ...- gst_format_logs.sh
#!/bin/bash
input_path=$1
awk -F'\\\\ ' 'BEGIN {OFS="\n";ORS="\n\n"} $1 ~ /^\/GstPipeline/ {$1=$1;print $0}' ${input_path}
exit 0- Debug:
sent to stderr (2)
-
- Debugging
tools
export GST_DEBUG=1 # default
export GST_DEBUG="*:2"
export GST_DEBUG=WARN,udpsrc:INFO,videodecoder:DEBUG
export GST_DEBUG=3,rtpjitterbuffer:3,rtpbasedepayload:6,videodecoder:4-
number
name
1
ERROR
2
WARNING
3
FIXME
4
INFO
5
DEBUG
6
LOG
7
TRACE
8
9
MEMDUMP
-
- Gstreamer pipeline diagram
-
- How to generate a Gstreamer pipeline diagram (graph)
- Dependencies
-
- Mageia
-
urpmi graphviz
- Utilització / Usage
-
mkdir /tmp/dots
export GST_DEBUG_DUMP_DOT_DIR=/tmp/dots
gst-launch ...cd /tmp/dots- to generate svg:
-
dot -Tsvg ...-gst-launch...PLAYING....dot >pipeline.svggwenview pipeline.svg
- to generate png:
-
dot -Tpng ...-gst-launch...PLAYING....dot >pipeline.pnggwenview pipeline.png
- Debugging
tools
- Sintaxi / Syntax
-
-
element:
ELEMENTTYPE [PROPERTY1 ...]
elements can be put into bins:
[BINTYPE.] ( [PROPERTY1...] PIPELINE-DESCRIPTION )
property:
NAME=*[(TYPE)]*VALUEin lists and ranges:*[(TYPE)]*VALUE
- range:
[VALUE,VALUE]
- list:
{VALUE[,VALUE...]}
-i int
-f float-4 fourcc-b bool boolean-s str string-fraction
link:
[[SRCELEMENT].[PAD1,...]] ! [[SINKELEMENT].[PAD1,...]] [[SRCELEMENT].[PAD1,...]] ! CAPS ! [[SINKELEMENT].[PAD1,...]]
caps:
MIMETYPE [,PROPERTY[,PROPERTY...]]] [;CAPS[;CAPS...]]
- range:
-
input
demuxer
decoder
filter
encoder muxer
demux
buffer
parse
(to get specific packets from demuxer)
decode
filter
encode codec parameters
(CAPS)
parse
(to prepare specific packets for muxer)mux
output
file
- filesrc location=videofile
- uri=file:///path/to/test.ts
devices
- dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 pids=111:112
network
unix
- fdsrc fd=...
- shmsrc
socket-path=...
is-live=...
- ... do-timestamp=...
decodebin name=decoder
(does not include sdpdemux)
- corresponding parameters (caps) can be grouped, but must appear after filter declaration (except videoconvert: videoconvert ! <caps> ! videoconvert)
- if a value in caps must be used,
the corresponding filter must be present. If the value in caps is not
used, because the input already has this value, the filter is not needed
filter
description
mimetype, comma separated key=value
video
video/x-raw videoscale
width=360 height=288
pixel-aspect-ratio=1/1
videorate
framerate=25/1 videoconvert
format=BGRA ?
interlace-mode=progressive
audio
audio/x-raw
audiorate
Drops/duplicates/adjusts timestamps on audio samples to make a perfect stream
tolerance=...
...
audioresample
Resamples audio
rate=48000
audioconvert Convert audio to different formats
format=S16LE
channels=2
layout=interleaved
- video
-
- x264enc
bitrate=512 (kb/s)
- omxh264enc target-bitrate=100000 control-rate=variable inline-header=true periodicty-idr=250
- vp8enc
target-bitrate=512000
(b/s)
- vp9enc
target-bitrate=512000
(b/s)
- x264enc
bitrate=512 (kb/s)
- audio
- video
-
- video/x-h264,
profile=baseline
- video/x-h264,
- video
-
- h264parse
- audio
-
- mpegaudioparse
- aacparse
- mpegtsmux
name=mux
- flvmux streamable=true name=mux
- mp4mux
faststart=true
file
- filesink location=music.ogg
devices
- autovideosink
- xvimagesink
- autoaudiosink
- omxanalogaudiosink
network
- udpsink host=192.168.0.8 port=5004 sync=false
- rtmpsink
location=rtmp://rtmp_server:1935/app/stream
unix
demux
specific stream (source Element Pads)
- demux.
matroskademux
- demux.audio_%u
- demux.video_%u
- demux.subtitle_%u
qtdemux name=demux - demux.video_0
- demux.audio_0
- demux.audio_1
- ...
sdpdemux name=demux
- demux.stream_0
- demux.stream_1
- ...
tsdemux program-number=805 name=demux - demux.audio_xxxx
flvdemux name=demux
- audio
- video
...
- queue [leaky=...]
- multiqueue
- video
-
- mpegvideoparse
- h264parse
- audio
-
- mpegaudioparse
- aacparse
- decodebin
- video
-
- omxmpeg2videodec
- omxh264dec
- audio
-
- avdec_mpeg2video
- video
-
- videotestsrc
[pattern=snow] [is-live=1] ...
- videotestsrc
[pattern=snow] [is-live=1] ...
- audio
-
- audiotestsrc [wave=5] [volume=0.8] [freq=200]
- video
-
- video/x-raw,
framerate=25/1, width=640, height=360,
- if format is specified,
videoconvert must be specified after it
-
- format=...,
...
- format=...,
- video/x-raw,
framerate=25/1, width=640, height=360,
- audio
-
- audio/x-raw,
format=...,
layout=...,
rate=...,
channels=...
- audio/x-raw,
-
-
sdpdemux
rtpbin
udpsrc
rtpsession
rtpssrcdemux
rtpjitterbuffer
rtpptdemux
- Sources and sinks
-
- GstBaseSrc
-
- do-timestamp
-
- it has to be specified at source if
we want lipsync at the output: sync=true
- it has to be specified at source if
we want lipsync at the output: sync=true
- GstBaseSink
- fd
(file
descriptor) (see Snowmix audio)
-
- fdsink
-
- ...
- ...
- fdsrc
-
- ...
- Exemple / Example
- send audio to file descriptor 3 and
listen to it. Verbose and warning logs are shown in terminal
export GST_DEBUG=WARN
AUDIOCAPS="audio/x-raw,format=S16LE,layout=interleaved,rate=44100,channels=2"
gst-launch-1.0 -v audiotestsrc wave=5 ! volume volume=0.1 ! ${AUDIOCAPS} ! fdsink fd=3 3>&1 1>&2 | gst-launch-1.0 fdsrc ! ${AUDIOCAPS} ! autoaudiosink
- shm
(shared memory) (see Snowmix video)
-
- shmsink
-
gst-launch-1.0 -v videotestsrc ! video/x-raw,framerate=25/1,width=640,height=480,format=BGRA! videoconvert !shmsink socket-path=/tmp/feed1 shm-size=`echo 640*480*4*22 | bc` wait-for-connection=0gst-launch-1.0 -vvideotestsrc is-live=true do-timestamp=true ! video/x-raw,framerate=25/1,width=640,height=480,format=BGRA ! videoconvert ! clockoverlay halignment=right valignment=top shaded-background=true font-desc="Sans, 24" ! shmsink socket-path=/tmp/feed1 shm-size=`echo 640*480*4*22 | bc` wait-for-connection=1 sync=true
gst-launch-1.0 filesrc location=sintel_timecode_640x272_44100_stereo.mp4 ! qtdemux name=demux demux. ! decodebin ! videoconvert ! videoscale ! videorate ! video/x-raw,width=320,height=136,format=BGRA ! shmsink socket-path=/tmp/feed1 shm-size=`echo 320*136*4*22 | bc -l` wait-for-connection=1 sync=true
- shmsrc
-
- you must specify:
width, height, framerate, format+videoconvert, and they shoud match values specified in shmsink
gst-launch-1.0 -v shmsrc socket-path=/tmp/feed1 do-timestamp=true is-live=true ! video/x-raw,width=640,height=480,framerate='25/1',format=BGRA ! videoconvert ! autovideosink
- to play at a framerate different from the input, specify a different framerate and add videorate
- you must specify:
du -h /dev/shm
ls -l /dev/shmnetstat -pena --unix
- shm (video) + fd (audio)
-
send
play
Notes
audio
feed_rate=44100
feed_channels=2
AUDIOCAPS="audio/x-raw,format=S16LE,layout=interleaved,rate=${feed_rate},channels=${feed_channels}"
gst-launch-1.0 \
audiotestsrc wave=5 ! volume volume=0.1 ! ${AUDIOCAPS} ! fdsink fd=3 3>&1 1>&2 \
| gst-launch-1.0 fdsrc ! ${AUDIOCAPS} ! autoaudiosink
video
ratefraction="25/1"
feed_width=320
feed_height=180
video_input_sar="1:1"
pixel_aspect_ratio=${video_input_sar/:/\/} # replace : -> /
VIDEOCAPS="video/x-raw,framerate=${ratefraction},width=${feed_width},height=${feed_height},pixel-aspect-ratio=${pixel_aspect_ratio}"
FORMAT_SHM="BGRA"
VIDEOCAPS_WITH_FORMAT="${VIDEOCAPS},format=${FORMAT_SHM}"
shm_socket_path=/tmp/shm_toto
FORMAT_DISPLAY="I420"
VIDEOCONVERT_DISPLAY="video/x-raw,format=${FORMAT_DISPLAY}"
FORMAT_SHM="BGRA"
VIDEOCONVERT_SHM="video/x-raw,format=${FORMAT_SHM}"
gst-launch-1.0 \
videotestsrc ! ${VIDEOCAPS_WITH_FORMAT} ! videoconvert ! shmsink socket-path=${shm_socket_path} shm-size=`echo ${feed_width}*${feed_height}*4*22 | bc` wait-for-connection=0 \# send video with tee to monitor
gst-launch-1.0 \
filesrc location=${video_path} ! qtdemux name=demux \
demux.video_0 ! queue ! decodebin ! videoscale method=5 ! queue ! videorate ! ${VIDEOCAPS} ! tee name=tv \
tv. ! queue ! videoconvert ! ${VIDEOCONVERT_SHM} ! shmsink socket-path=${shm_socket_path} shm-size=`echo ${feed_width}*${feed_height}*4*22 | bc` wait-for-connection=0 \
tv. ! queue ! videoconvert ! ${VIDEOCONVERT_DISPLAY} ! autovideosink sync=true \
3>&1 1>&2
gst-launch-1.0 shmsrc socket-path=${shm_socket_path} do-timestamp=true is-live=true ! ${VIDEOCAPS_WITH_FORMAT} ! videoconvert ! autovideosink
audio + video
gst-launch-1.0 \
videotestsrc ! ${VIDEOCAPS_WITH_FORMAT} ! videoconvert ! shmsink socket-path=${shm_socket_path} shm-size=`echo ${feed_width}*${feed_height}*4*22 | bc` wait-for-connection=0 \
audiotestsrc wave=5 ! volume volume=0.1 ! ${AUDIOCAPS} ! fdsink fd=3 3>&1 1>&2 \gst-launch-1.0 \\
filesrc location=${video_path} ! qtdemux name=demux \
demux.video_0 ! queue ! decodebin ! videoscale method=5 ! queue ! videorate ! videoconvert ! ${VIDEOCAPS_WITH_FORMAT} ! \
shmsink socket-path=${shm_socket_path} shm-size=`echo ${feed_width}*${feed_height}*4*22 | bc` wait-for-connection=0 \
demux.audio_0 ! queue ! decodebin ! audioconvert ! audioresample ! queue ! audiorate ! $AUDIOCAPS ! tee name=ta \
ta. ! queue ! fdsink fd=3 \
3>&1 1>&2# send audio and video with tee to monitor
gst-launch-1.0 \
filesrc location=${video_path} ! qtdemux name=demux \
demux.video_0 ! queue ! decodebin ! videoscale method=5 ! queue ! videorate ! ${VIDEOCAPS} ! tee name=tv \
tv. ! queue ! videoconvert ! ${VIDEOCONVERT_SHM} ! shmsink socket-path=${shm_socket_path} shm-size=`echo ${feed_width}*${feed_height}*4*22 | bc` wait-for-connection=0 \
tv. ! queue ! videoconvert ! ${VIDEOCONVERT_DISPLAY} ! autovideosink sync=true \
demux.audio_0 ! queue ! decodebin ! audioconvert ! audioresample ! queue ! audiorate ! ${AUDIOCAPS} ! tee name=ta \
ta. ! queue ! autoaudiosink sync=true \
ta. ! queue ! fdsink fd=3 \
3>&1 1>&2 \
| gst-launch-1.0 \
shmsrc socket-path=${shm_socket_path} do-timestamp=true is-live=true ! ${VIDEOCAPS_WITH_FORMAT} ! videoconvert ! autovideosink \
fdsrc ! ${AUDIOCAPS} ! autoaudiosink
- es queden fitxers creats (/tmp/shm_toto, /dev/shm/shmpipe*) quan:
- no arrenca
- s'atura amb la X de la finestra de la sortida
- s'esborren bé els fitxers creats quan:
- s'atura amb CTRL-C
-
- Demux
-
- sdpdemux
-
- Play from SDP file
- Includes
- Parameters
-
latency(ms)
-
INFO rtpjitterbuffer gstrtpjitterbuffer.c:3942:do_deadline_timeout:<rtpjitterbuffer0> got deadline timeout
- Codecs
-
- x264enc
-
aud=false-
- do not generate NALs of type 9 (Access Unit Delimiter)
(they generate problems when playing WebRTC on Chrome)
- do not generate NALs of type 9 (Access Unit Delimiter)
(they generate problems when playing WebRTC on Chrome)
- Bins
-
-
playbinuridecodebin typefind queue2 decodebin typefind - flvdemux
multiqueue
- aacparse
- avdec_aac
- h264parse
capsfilter
- avdec_h264
input-selector playsink
tee
streamsynchronizer
bin
queue
playsinkaudioconvert
audioconvert
audioresample
- pulsesink
bin
videoconvert
deinterlace
bin
queue
playsinkvideoconvert
videoconvert
videoscale
videobalance
videoconvert
- xvimagesink
input-selector - flvdemux
- decodebin3
- rtpbin
-
- GStreamer pipeline recipe: Stream file via RTP using rtpbin (gst-launch)
- How
te retrieve stream statistics in Gstreamer? (Python)
- Includes
-
- rtpsession
-
- rtcp-min-interval
- rtcp-min-interval
- rtpssrcdemux
- rtpjitterbuffer
- rtpptdemux
-
rtpbin
Examples
sink
src
transmission
video
rtppay
send_rtp_sink_0
send_rtp_src_0
udpsink
send_rtcp_src_0
udpsink sync=false async=false (SR)
audio
rtppay
send_rtp_sink_1
send_rtp_src_1
udpsink
send_rtcp_src_1
udpsink sync=false async=false (SR)
reception
video
udpsrc
recv_rtp_sink_0
recv_rtp_src_0_<ssrc_0>_<pt_0>
rtpdepay
udprsc (SR)
recv_rtcp_sink_0
send_rtcp_src_0
udpsinksync=false async=false(RR)
audio
udpsrc
recv_rtp_sink_1
recv_rtp_src_1_<ssrc_1>_<pt_1>
rtpdepay
udpsrc (SR)
recv_rtcp_sink_1
send_rtcp_src_1
udpsinksync=false async=false(RR)
-
- Play
-
- general: using
playbin -
gst-launch-1.0 -v playbin uri=...
-
gst-launch-1.0 -v playbin uri=...
gst-launch-1.0 uridecodebin uri=... name=decoder
playsink name=sink decoder.src_0 ! sink.video_sink decoder.src_1 ! sink.audio_sink
- from testsrc
-
- videotestsrc
-
gst-launch-1.0 -v videotestsrc! video/x-raw,framerate=25/1,width=1280,height=720 ! autovideosinkgst-launch-1.0 -v videotestsrc pattern=snow ! video/x-raw,framerate=12/1,width=1280,height=720 ! autovideosinkgst-launch-1.0 -v videotestsrc ! video/x-raw,framerate=12/1,width=1280,height=720,format=BGRA ! videoconvert ! autovideosink- clock overlay
- time overlay
-
gst-launch-1.0 -v videotestsrc is-live=true ! timecodestamper ! timeoverlay shaded-background=true 'time-mode=time-code' font-desc="Sans, 24" ! autovideosinkgst-launch-1.0 -v videotestsrc is-live=true ! video/x-raw, framerate=25/1, width=640, height=360 ! timecodestamper ! timeoverlay halignment=right valignment=bottom text="Stream time:" shaded-background=true font-desc="Sans, 24" ! autovideosink
- clock + time overlay
-
gst-launch-1.0 -v videotestsrc is-live=true ! video/x-raw, framerate=25/1, width=640, height=360 ! timecodestamper ! timeoverlay halignment=left valignment=top shaded-background=true font-desc="Sans, 24" ! clockoverlay halignment=right valignment=top shaded-background=true font-desc="Sans, 24" ! autovideosink
- audiotestsrc
-
gst-launch-1.0 -v audiotestsrc ! autoaudiosink- white noise (
wave=5), stereo (channels=2)
-
gst-launch-1.0 -v audiotestsrc is-live=true wave=5 ! 'audio/x-raw,format=S16LE,layout=interleaved,rate=48000,channels=2' ! autoaudiosink
- test video + audio
-
gst-launch-1.0 -v videotestsrc ! video/x-raw,framerate=25/1,width=1280,height=720 ! autovideosink audiotestsrc ! autoaudiosink
- from DVB device
-
- only video from DVB device (mpegvideoparse is needed because if not, maybe teletext is taken; and error is shown):
-
gst-launch-1.0 -v dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=805 ! queue ! mpegvideoparse ! decodebin ! autovideosink
- only audio:
-
gst-launch-1.0 -v dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=805 ! queue ! mpegaudioparse ! decodebin ! autoaudiosinkgst-launch-1.0 dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux demux. ! queue ! mpegaudioparse ! decodebin ! omxanalogaudiosink
- audio and video from program 805 in DVB input:
-
gst-launch-1.0 -v dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=805 name="demux" \
demux. ! queue ! mpegaudioparse ! decodebin ! autoaudiosink \
demux. ! queue ! mpegvideoparse ! decodebin ! autovideosink
- from file
-
gst-launch-1.0 -v playbin uri=file:/absoulte/path/to/your_video_filegst-launch-1.0 \
filesrc location=${video_path} ! decodebin name=dec \
dec. ! queue ! autovideosink \
dec. ! queue ! autoaudiosink
- MP4
-
gst-launch-1.0 -v playbinuri=file:/absoulte/path/to/toto.mp4- audio and video from an MP4 file (queue is needed when
playin audio and video)
-
gst-launch-1.0 filesrc location=sintel-1024-stereo.mp4 ! qtdemux name=demux \
demux. ! queue ! decodebin ! autovideosink \
demux. ! queue ! decodebin ! autoaudiosink
- rescale an anamorphic video
gst-launch-1.0 filesrc location=toto_720x576_anamorphic.mp4 ! qtdemux name=demux demux. ! queue ! decodebin ! videoscale ! video/x-raw,width=176,height=140,pixel-aspect-ratio=64/45 ! autovideosink
- only audio from an MP4 file
-
gst-launch-1.0 filesrc location=sintel-1024-stereo.mp4 ! qtdemux name=demux \
demux.audio_0 ! decodebin ! autoaudiosink
- TS
-
- program in a TS file (first program
found?):
-
gst-launch-1.0 -v playbinuri=file:/tmp/toto.ts
- only video from TS file
(program_number=802)
-
gst-launch-1.0 -v filesrc location=/disc/videos/tvc/tvc_794_20140821_1709.ts ! tsdemux program-number=802 ! mpegvideoparse ! decodebin ! autovideosink
- program in a TS file (first program
found?):
- OGG
-
- audio and video from an OGG file (queue is needed when
playin audio and video)
-
gst-launch-1.0 filesrc location=sintel_trailer-720p.ogv ! oggdemux name=demux \
demux. ! queue ! decodebin ! autovideosink \
demux. ! queue ! decodebin ! autoaudiosink
- audio and video from an OGG file (queue is needed when
playin audio and video)
- WebM / Matroska
-
gst-launch-1.0 -v playbinuri=file:/absoulte/path/to/toto.webm- playbinaudio and video from a webm file:
-
gst-launch-1.0 -v \
filesrc location=/path/to/toto.webm ! matroskademux name=demux \
demux.video_0 ! queue ! decodebin ! autovideosink sync=true \
demux.audio_0 ! queue ! decodebin ! autoaudiosink sync=true
gst-launch-1.0 -v \
filesrc location=/path/to/toto.webm ! matroskademux name=demux \
demux. ! queue ! vp8dec ! autovideosink sync=true \
demux. ! queue ! opusdec ! autoaudiosink sync=true
- SDP
-
- See also: play from RTP
- sdpdemux
- UDP buffer
-
- value is taken from kernel
parameter net.core.rmem_default
GST_DEBUG=3,udpsrc:4 gst-launchfilesrc location=toto.sdp ! sdpdemux name=demux ...-
udpsrc gstudpsrc.c:1428:gst_udpsrc_open:<udpsrc0> have udp buffer of 212992 bytes
- value is taken from kernel
parameter net.core.rmem_default
- audio and video from SDP file (RTP):
-
gst-launch-1.0 filesrc location=toto.sdp do-timestamp=true ! sdpdemux latency=1000 debug=true name=bin \
bin. ! "application/x-rtp, media=(string)video" ! decodebin ! autovideosink sync=true \
bin. ! "application/x-rtp, media=(string)audio" ! decodebin ! autoaudiosink sync=true
# RTP + RTCP, using sdpdemux
gst-launch-1.0 \
filesrc location=$sdp_path do-timestamp=true ! sdpdemux latency=${sdpdemux_latency_ms} name=bin \
bin. ! "application/x-rtp, media=(string)video" ! queue ! decodebin ! videoconvert ! videoscale ! videorate ! autovideosink sync=true \
bin. ! "application/x-rtp, media=(string)audio" ! queue ! decodebin ! audioconvert ! audioresample ! audiorate ! autoaudiosink sync=true
- H.264 + AAC
-
gst-launch-1.0 filesrc location=toto.sdp do-timestamp=true ! sdpdemux name=demux \
demux. ! queue ! rtph264depay ! decodebin ! autovideosink sync=true \
demux. ! queue ! rtpmp4gdepay ! decodebin ! autoaudiosink sync=true
- only video from SDP file (RTP). Make sure that video is the first stream specified in sdp file:
-
gst-launch-1.0 filesrc location=toto.sdp ! sdpdemux name=demux \
demux.stream_0 ! queue ! decodebin ! autovideosink
- only audio from SDP file (RTP). Make sure that audio is the second stream specified in sdp file:
-
gst-launch-1.0 filesrc location=toto.sdp ! sdpdemux name=demux \
demux.stream_1 ! queue ! decodebin ! autoaudiosink
- Problemes / Problems
-
- Pèrdua
de paquets / Packet loss
-
- Pèrdua de
paquets amb ffmpeg / Packet loss with ffmpeg
- Frame drop when playing RTP stream using SDP file
- Solució / Solution
-
- Augmenteu la mida del buffer de recepció UDP modificant el paràmetre del kernel parameter net.core.rmem / Increase UDP reception buffer size by modifying kernel parameter net.core.rmem:
-
sudo sysctl -w net.core.rmem_default=851968
- potser també haureu d'augmentar el màxim / you may need to also increase the maximum:
-
sudo sysctl -w net.core.rmem_max=851968
export GST_DEBUG=WARN,udpsrc:INFO; gst-launch-1.0 ...
-
udpsrc gstudpsrc.c:1428:gst_udpsrc_open:<udpsrc3> have udp buffer of 851968 bytes
- Pèrdua de
paquets amb ffmpeg / Packet loss with ffmpeg
- Manca de fluïdesa / Lack of smoothness
-
videodecoder gstvideodecoder.c:2775:gst_video_decoder_prepare_finish_frame:<avdec_h264-0> decreasing timestamp (0:00:00.008558259 < 0:00:00.058900688)-
- Solució / Solution
-
- Increase latency parameter (default: 200 ms) for sdpdemux:
gst-launch-1.0 filesrc location=toto.sdp ! sdpdemux latency=400 name=demux ...
audiobasesink gstaudiobasesink.c:1787:gst_audio_base_sink_get_alignment:<autoaudiosink0-actual-sink-alsa> Unexpected discontinuity in audio timestamps of +0:00:00.131360544, resyncing-
- Solució / Solution
-
- ...
- ...
- “delayed linking failed”
-
- Gstreamer
“delayed linking failed” with rtspsrc video and audio
- Solució / Solution
-
... "application/x-rtp, media=(string)audio" ! decodebin ...... "application/x-rtp, media=(string)video" ! decodebin ...
- Gstreamer
“delayed linking failed” with rtspsrc video and audio
- lipsync
- Pèrdua
de paquets / Packet loss
- See also: play from RTP
- from network
-
- RTP
-
- See also: Play from SDP file
- Problemes / Problems
-
- udpsrc
-
videodecoder gstvideodecoder.c:2775:gst_video_decoder_prepare_finish_frame:<avdec_h264-0> decreasing timestamp (0:00:45.183879818 < 0:00:45.188331700)
gst-launch-1.0 udpsrc address=127.0.0.1 port=5004 ! "application/x-rtp, media=(string)video, clock-rate=(int)90000, encoding-name=(string)H264, packetization-mode=(string)1, profile-level-id=(string)64001f, payload=(int)96" ! queue ! \
rtph264depay ! decodebin ! autovideosinkgst-launch-1.0 -v \
udpsrc address=${address} port=${video_port} do-timestamp=true ! queue ! "application/x-rtp, media=(string)video, payload=(int)96, clock-rate=(int)90000, encoding-name=(string)H264, packetization-mode=(string)1, sprop-parameter-sets=(string)\"Z2QAHqzZQKAv+XARAAADAAEAAAMAPA8WLZY\=\,aOvssiw\=\", profile-level-id=(string)64001E" ! rtph264depay ! decodebin ! autovideosink sync=true \
udpsrc address=${address} port=${audio_port} do-timestamp=true ! queue ! "application/x-rtp, media=(string)audio, payload=(int)97, clock-rate=(int)44100, encoding-name=(string)MPEG4-GENERIC, encoding-params=(string)2, profile-level-id=(string)1, mode=(string)AAC-hbr, sizelength=(string)13, indexlength=(string)3, indexdeltalength=(string)3, config=(string)121056E500" ! rtpmp4gdepay ! decodebin ! autoaudiosink sync=true-
- caps can be obtained e.g. by
executing:
gst-launch -v ... sdpdemux ...
- caps can be obtained e.g. by
executing:
- common variables:
-
# ffmpeg -re -i easylife.mp4 -c:v copy -an -f rtp -cname toto rtp://234.1.2.3:5004 -vn -c:audio copy -f rtp -cname toto rtp://234.1.2.3:5006 -sdp_file /mnt/nfs/sdp/toto.sdp
address=234.1.2.3
video_rtp_port=5004
video_rtcp_port=$(( video_rtp_port + 1 ))
VIDEOCAPS="application/x-rtp, media=(string)video, payload=(int)96, clock-rate=(int)90000, encoding-name=(string)H264, packetization-mode=(string)1, sprop-parameter-sets=(string)\"Z2QAHqzZQKAv+XARAAADAAEAAAMAPA8WLZY\=\,aOvssiw\=\", profile-level-id=(string)64001E"
audio_rtp_port=$(( video_rtp_port + 2 ))
audio_rtcp_port=$(( video_rtp_port + 3 ))
AUDIOCAPS="application/x-rtp, media=(string)audio, payload=(int)97, clock-rate=(int)44100, encoding-name=(string)MPEG4-GENERIC, encoding-params=(string)2, profile-level-id=(string)1, mode=(string)AAC-hbr, sizelength=(string)13, indexlength=(string)3, indexdeltalength=(string)3, config=(string)121056E500"
- RTP
-
# RTP
gst-launch-1.0 \
udpsrc address=${address} port=${video_rtp_port} do-timestamp=true ! "$VIDEOCAPS" ! \
rtph264depay ! queue ! decodebin ! autovideosink sync=true \
udpsrc address=${address} port=${audio_rtp_port} do-timestamp=true ! "$AUDIOCAPS" ! \
rtpmp4gdepay ! queue ! decodebin ! autoaudiosink sync=true
- RTP (using rtpbin)
-
# RTP, using rtpbin
gst-launch-1.0 \
rtpbin name=bin \
udpsrc address=${address} port=${video_rtp_port} do-timestamp=true ! "$VIDEOCAPS" ! bin.recv_rtp_sink_0 \
bin. ! rtph264depay ! queue ! decodebin ! autovideosink sync=true \
udpsrc address=${address} port=${audio_rtp_port} do-timestamp=true ! "$AUDIOCAPS" ! bin.recv_rtp_sink_1 \
bin. ! rtpmp4gdepay ! queue ! decodebin ! autoaudiosink sync=true
- RTP with RTCP (using rtpbin)
-
# RTP + RTCP, using rtpbin
gst-launch-1.0 \
rtpbin name=bin \
udpsrc address=${address} port=${video_rtp_port} do-timestamp=true ! "$VIDEOCAPS" ! bin.recv_rtp_sink_0 \
bin. ! "application/x-rtp, media=(string)video" ! queue ! decodebin ! autovideosink sync=true \
udpsrc address=${address} port=${video_rtcp_port} ! "application/x-rtcp" ! bin.recv_rtcp_sink_0 \
udpsrc address=${address} port=${audio_rtp_port} do-timestamp=true ! "$AUDIOCAPS" ! bin.recv_rtp_sink_1 \
bin. ! "application/x-rtp, media=(string)audio" ! queue ! decodebin ! autoaudiosink sync=true \
udpsrc address=${address} port=${audio_rtcp_port} ! "application/x-rtcp" ! bin.recv_rtcp_sink_1
# RTP + RTCP, using rtpbin
gst-launch-1.0 \
rtpbin name=bin \
udpsrc address=${address} port=${video_rtp_port} do-timestamp=true ! "$VIDEOCAPS" ! bin.recv_rtp_sink_0 \
bin. ! rtph264depay ! queue ! decodebin ! autovideosink sync=true \
udpsrc address=${address} port=${video_rtcp_port} ! "application/x-rtcp" ! bin.recv_rtcp_sink_0 \
udpsrc address=${address} port=${audio_rtp_port} do-timestamp=true ! "$AUDIOCAPS" ! bin.recv_rtp_sink_1 \
bin. ! rtpmp4gdepay ! queue ! decodebin ! autoaudiosink sync=true \
udpsrc address=${address} port=${audio_rtcp_port} ! "application/x-rtcp" ! bin.recv_rtcp_sink_1
- Problemes / Problems
-
- lipsync
-
- source: specify do-timestamp=true
- sinks: specify sync=true
- sdpdemux: latency must be big enough. If stream has B images, try with latency=1000
- check that RTP flow contains RTCP packets with "Source Description" (SDES) information
-
- ffmpeg does not
generate "Source Description" in RTCP, by default. Force it by
adding option -cname with the same value in both audio and video
streams:
ffmeg -an ... -f rtp -cname user123456789@host-87654321 rtp://... ... -vn ... -f rtp -cname user123456789@host-87654321 rtp://...
- ffmpeg does not
generate "Source Description" in RTCP, by default. Force it by
adding option -cname with the same value in both audio and video
streams:
- HTTP
-
- check that souphttpsrc is present
-
gst-inspect-1.0 | grep souphttpsrc
- if not present, compile it
-
- Dependencies
-
- CentOS
-
sudo yum install libsoup-devel
- Mageia
-
urpmi libsoup-devel
- gst-plugins-good
-
./configuremakesudo make install
gst-launch-1.0playbinuri=http://download.blender.org/peach/bigbuckbunny_movies/BigBuckBunny_320x180.mp4
- RTMP
-
gst-launch-1.0 -vplaybinuri=rtmp://nginx-server/myapp/mystream
gst-launch-1.0 -v \
rtmpsrc location=${source} do-timestamp=true ! queue2 ! decodebin name=mydecoder \
mydecoder. ! autovideosink sync=true \
mydecoder. ! autoaudiosink sync=true
(not working?) source=rtmp://nginx-server/myapp/mystream
gst-launch-1.0 \
rtmpsrc location=${source} do-timestamp=true ! flvdemux name=demux \
demux.video ! queue ! decodebin ! autovideosink sync=true \
demux.audio ! queue ! decodebin ! autoaudiosink sync=true
- general: using
- Mux to
-
- test to TS
-
gst-launch-1.0 -v videotestsrc ! video/x-raw,framerate=24/1,width=1280,height=720 ! videoconvert ! x264enc ! video/x-h264,profile=high ! mpegtsmux ! filesink location=toto.ts
- test to MP4
-
- NOTE:
stream-format=(string)byte-streamis not supported by MP4 gst-launch-1.0 -v videotestsrc ! video/x-raw,framerate=24/1,width=1280,height=720 ! videoconvert ! x264enc ! video/x-h264,profile=high ! mp4mux ! filesink location=toto.mp4-
- Problem:
-
- moov atom not found
-
- Solution
-
- (?) mp4mux
faststart=true
- (?) mp4mux
faststart=true
- NOTE:
- Transmux
-
- rtmp -> rtmp
gst-launch-1.0 rtmpsrc location=rtmp://server.org/my_app/first do-timestamp=true ! queue2 ! flvdemux name=demux\
flvmux name=mux\
demux.video ! queue ! mux.video\
demux.audio ! queue ! mux.audio\
mux.src ! queue ! rtmpsink location=rtmp://server.org/my_app/second
- Note: if using nginx-rtmp-module as destination, check nginx.conf configuration
- only video (mp4 -> ts):
-
gst-launch-1.0 filesrc location=sintel-1024-stereo.mp4 ! qtdemux name=demux \
mpegtsmux name=mux ! filesink location=toto.ts \
demux. ! queue ! h264parse ! mux.gst-launch-1.0 filesrc location=sintel-1024-stereo.mp4 ! qtdemux name=demux \
mpegtsmux name=mux ! filesink location=toto.ts \
demux. ! video/x-h264 ! queue ! h264parse ! mux.
- video and audio (mp4 -> ts)
-
gst-launch-1.0 filesrc location=sintel-1024-stereo.mp4 ! qtdemux name=demux \
mpegtsmux name=mux ! filesink location=toto2015.ts \
demux. ! queue ! h264parse ! mux. \
demux. ! queue ! aacparse ! mux.
- video and audio (mp4 -> flv)
-
gst-launch-1.0 -v filesrc location=sintel-1024-stereo.mp4 ! qtdemux name=demux \
flvmux streamable=true name=mux ! filesink location=toto.flv \
demux. ! queue ! h264parse ! mux. \
demux. ! queue ! aacparse ! mux.
- video and audio (flv ->mp4)
-
- ...
- ...
- video (H.264) (sdp -> ts)
-
gst-launch-1.0 -v filesrc location=toto.sdp ! sdpdemux name=demux \
mpegtsmux name=mux ! filesink location=toto.ts \
demux. ! queue ! rtph264depay ! mux.
- video (H.264) and audio (AAC) (sdp -> ts)
-
gst-launch-1.0 -v filesrc location=toto.sdp ! sdpdemux name=demux \
mpegtsmux name=mux ! filesink location=toto.ts \
demux. ! queue ! rtph264depay ! mux. \
demux. ! queue ! rtpmp4gdepay ! mux.
- video (sdp->mp4) (not working)
-
gst-launch-1.0 filesrc location=/tmp/bbb.sdp ! sdpdemux name=demux \
mp4mux name=mux ! filesink location=/tmp/toto.mp4 \
demux. ! rtph264depay ! h264parse ! mux.
-
- problem
-
ffplay toto.mp4
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f128c0008c0] moov atom not found
toto.mp4: Invalid data found when processing input- Solution?
-
- mp4mux faststart=true
- mp4mux faststart=true
- video and audio (sdp -> mp4) (not
working)
-
gst-launch-1.0 -v filesrc location=toto.sdp ! sdpdemux name=demux \
mp4mux name=mux ! filesink location=toto.mp4 \
demux. ! queue ! h264parse ! mux. \
demux. ! queue ! aacparse ! mux.
- Transcode
-
- only video, to file:
-
gst-launch-1.0 -v dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=805 name="demux" \
demux. ! queue ! mpegvideoparse ! decodebin ! videoconvert ! x264enc ! video/x-h264,stream-format=byte-stream,profile=high ! h264parse ! \
mpegtsmux ! filesink location=/tmp/toto.ts
- and resize, stream:
-
gst-launch-1.0 -v filesrc location=tvc_20150604.ts ! tsdemux program-number=806 ! \
mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=320, height=320 ! videoconvert ! omxh264enc ! h264parse ! \
mpegtsmux ! udpsink host=192.168.0.8 port=5004 sync=falsegst-launch-1.0 -v dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 ! \
mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=320, height=320 ! videoconvert ! omxh264enc ! h264parse ! \
mpegtsmux ! udpsink host=192.168.0.8 port=5004 sync=false
gst-launch-1.0 -v dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=805 name="demux" \
demux. ! queue ! mpegvideoparse ! decodebin ! videoscale ! 'video/x-raw, width=360, height=288' ! videoconvert ! x264enc ! video/x-h264,stream-format=byte-stream,profile=main ! h264parse ! \
mpegtsmux ! udpsink host=192.168.0.8 port=5004 sync=false
- only video, to file:
- tee
- two windows from MP4 file:
gst-launch-1.0 \
filesrc location=${video_path} ! qtdemux name=demux \
demux. ! queue ! decodebin ! tee name=tv \
tv. ! queue ! autovideosink sync=true \
tv. ! queue ! autovideosink sync=true \
demux. ! queue ! decodebin ! tee name=ta \
ta. ! queue ! autoaudiosink sync=true \
ta. ! queue ! autoaudiosink sync=true
- two windows from SDP file:
gst-launch-1.0 \
filesrc location=${video_path} do-timestamp=true ! sdpdemux name=bin \
bin. ! "application/x-rtp, media=(string)audio" ! queue ! decodebin ! tee name=ta \
ta. ! queue ! autoaudiosink sync=true \
ta. ! queue ! autoaudiosink sync=true \
bin. ! "application/x-rtp, media=(string)video" ! queue ! decodebin ! tee name=tv \
tv. ! queue ! autovideosink sync=true \
tv. ! queue ! autovideosink sync=true
- Stream
-
- Introduction
to network streaming using GStreamer
- TS over UDP
-
- UDP unicast stream only audio:
-
gst-launch-1.0 dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux \
demux. ! queue ! mpegaudioparse ! decodebin ! audioconvert ! lamemp3enc ! \
mpegtsmux ! udpsink host=192.168.0.8 port=5004 sync=falsegst-launch-1.0 dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux \
demux. ! queue ! mpegaudioparse ! decodebin ! audioconvert ! lamemp3enc ! mux. \
mpegtsmux name=mux ! udpsink host=192.168.0.8 port=5004 sync=falsegst-launch-1.0 dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux \
mpegtsmux name=mux ! udpsink host=192.168.0.8 port=5004 sync=false \
demux. ! queue ! mpegaudioparse ! decodebin ! audioconvert ! lamemp3enc ! mux.
- Encode to H.264, mux to TS, UDP stream:
-
gst-launch-1.0 -e videotestsrc ! video/x-raw, framerate=25/1, width=640, height=360 ! x264enc ! \
mpegtsmux ! udpsink host=192.168.0.8 port=5004 sync=false
gst-launch-1.0 -v -e videotestsrc ! video/x-raw, framerate=25/1, width=640, height=360 ! x264enc bitrate=512 ! video/x-h264,profile=high ! h264parse ! \
mpegtsmux ! udpsink host=192.168.0.8 port=5004 sync=falsegst-launch-1.0 -e mpegtsmux name="muxer" ! udpsink host=192.168.0.8 port=5004 sync=false \
videotestsrc ! video/x-raw, framerate=25/1, width=640, height=360 ! x264enc bitrate=512 ! video/x-h264,profile=high ! h264parse ! muxer.
- Mux video and audio, UDP stream:
-
gst-launch-1.0 -e mpegtsmux name="muxer" ! udpsink host=192.168.0.8 port=5004 sync=false \
videotestsrc ! video/x-raw, framerate=25/1, width=640, height=360 ! x264enc bitrate=512 ! video/x-h264,profile=high ! h264parse ! muxer. \
audiotestsrc wave=5 ! audioconvert ! lamemp3enc ! muxer.gst-launch-1.0 dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux \
mpegtsmux name=mux ! udpsink host=192.168.0.8 port=5004 sync=false \
demux. ! queue ! mpegaudioparse ! decodebin ! audioconvert ! lamemp3enc ! mux. \
demux. ! queue ! mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=320, height=320 ! videoconvert ! omxh264enc inline-header=true periodicty-idr=50 ! h264parse ! mux.- Notes:
-
- default for omxh264enc is inline-header=true
- specification of periodicty-idr is needed by vlc to be played
- these parameters are available on latest version of gstreamer. Maybe you need to compile version 1.2.
- to RTP
-
- RTP
and RTSP support
- gstreamer/gst-plugins-good/gst/rtp/README
- Streaming
H.264 via RTP
- Play from SDP file
- SDP generation
-
- RTP components
- GStreamer SDP library
- Structure
GstSdp.SDPMessage- gst-plugins-base / gst-libs / gst / sdp
- webrtcbidirectional.c
- gstwebrtcbin.h
-
media
caps
SDP ret, media = GstSdp.SDPMedia.new()
caps = pad.get_current_caps()
(all sets)
ret = GstSdp.SDPMedia.set_media_from_caps(caps, media)
media.as_text()
- media.set_port_info(5004, 1)
m=<caps.media> <media.port> <media.proto> <caps.payload>
m=video 5004 RTP/AVP 96
- media.set_information("my info\n")
i=<media.information>
i=my info
- media.add_connection("IN",
"IP4", "1.2.3.4", 16, 1)
c=<media.connection.nettype> <media.connection.addrtype> <media.connection.address>
c=IN IP4 1.2.3.4
a=rtpmap:<caps.payload> <caps.encoding-name>/<caps.clock-rate>
a=rtpmap:96 H264/90000
(caps) a=fmtp:<caps.payload> ...
a=fmtp:96 ...
- mp4rtp.py
class Streamer(object):
def build_sdp(self, rtpbin, address, port, sdp_path, session_name=None):
"""
rtpbin: Gst.RtpBin element
address: destination address
port: initial destination port
sdp_path: created file with SDP
session_name: (optional) session name (s=...)
"""
ret, sdp_message = GstSdp.SDPMessage.new()
sdp_message.set_version('0')
ttl = 64
number_addresses = 1
sdp_message.set_connection("IN", "IP4", address, ttl, number_addresses)
if session_name:
sdp_message.set_session_name(session_name)
pads = rtpbin.iterate_pads()
while True:
ret, pad = pads.next()
if ret==Gst.IteratorResult.OK:
# only source pads
if pad.direction != Gst.PadDirection.SRC:
continue
# only pads with name send_rtp_src...
pad_name = pad.get_name()
if not pad_name.startswith('send_rtp_src'):
continue
print("pad: {0:s}".format(pad.name))
caps = pad.get_current_caps()
print(" {0:s}".format(caps.to_string()))
ret, media = GstSdp.SDPMedia.new()
if ret != GstSdp.SDPResult.OK:
print("Error")
return
media.set_port_info(port, 1)
port = port + 2
media.proto = "RTP/AVP"
ret = GstSdp.SDPMedia.set_media_from_caps(caps, media)
if ret != GstSdp.SDPResult.OK:
print("Error")
sdp_message.add_media(media)
elif ret==Gst.IteratorResult.DONE:
break
elif ret==Gst.IteratorResult.ERROR:
break
print(sdp_message.as_text())
f = open(sdp_path, 'w')
f.write(sdp_message.as_text())
def on_message(self, bus, msg, user_data):
t = msg.type
...
elif t == Gst.MessageType.STATE_CHANGED:
old, new, pending = msg.parse_state_changed()
if new == Gst.State.PAUSED:
if msg.src.name=="bin":
print("RtpBin")
rtpbin = msg.src
self.build_sdp(rtpbin, self.dst_address, self.initial_port, "output.sdp", None)
...
- webrtcbin
- ...
- Bug 795404 - webrtcbin: h.264 streams work with Chrome but not with Firefox
- "Inserting an fmtp attribute with a profile-level-id value seems to fix it."
- webrtc-test.cpp
- webrtc-unidirectional-h264.c
- Generating a SDP file from a streaming pipeline
- caps
to SDP (README)
-
gstreamer
sdp file
command
gst-launch-1.0 -v videotestsrc ! videoconvert ! x264enc ! rtph264pay config-interval=10 pt=96 ! udpsink host=234.1.2.3 port=5004
v=0
m=<media> <port> RTP/AVP <payload>
c=IN IP4 <host>
a=rtpmap:<payload> <encoding-name>/<clock-rate>
a=fmtp:96 packetization-mode=<packetization-mode>; sprop-parameter-sets=<sprop-parameter-sets>; profile-level-id=<profile-level-id>
caps
(given by -v)
application/x-rtp,
media=(string)video,
clock-rate=(int)90000,
encoding-name=(string)H264,
packetization-mode=(string)1,
profile-level-id=(string)f4000d,
sprop-parameter-sets=(string)"Z/QADZGbKCg/YC1BgEFQAAADABAAAAMDyPFCmWA\=\,aOvsRIRA",
payload=(int)96,
ssrc=(uint)3934427744,
timestamp-offset=(uint)2187273080,
seqnum-offset=(uint)1602,
a-framerate=(string)30
media=(string)audio,
...
-
send
receive
from file
common code
input_path=$1
sdp_path=/tmp/toto.sdp
dst_address=234.1.2.3
video_rtp_port=5004
video_rtcp_port=$(( video_rtp_port + 1 ))
video_media_subtype="H264"
rtp_video_payload_type=96
audio_rtp_port=$(( video_rtp_port + 2 ))
audio_rtcp_port=$(( video_rtp_port + 3 ))
audio_media_subtype="aac"
rtp_audio_payload_type=$(( rtp_video_payload_type + 1 ))
audio_rate=48000
audio_channels=2
sdp_path=$1
address=234.1.2.3
video_rtp_port=5004
video_rtcp_port=$(( video_rtp_port + 1 ))
rtp_video_payload_type=96
VIDEOCAPS="application/x-rtp, media=(string)video, payload=(int)96, clock-rate=(int)90000, encoding-name=(string)H264, packetization-mode=(string)1, sprop-parameter-sets=(string)\"Z2QAHqzZQKAv+XARAAADAAEAAAMAPA8WLZY\=\,aOvssiw\=\", profile-level-id=(string)64001E"
audio_rtp_port=$(( video_rtp_port + 2 ))
audio_rtcp_port=$(( video_rtp_port + 3 ))
rtp_audio_payload_type=$(( rtp_video_payload_type + 1 ))
AUDIOCAPS="application/x-rtp, media=(string)audio, payload=(int)97, clock-rate=(int)44100, encoding-name=(string)MPEG4-GENERIC, encoding-params=(string)2, profile-level-id=(string)1, mode=(string)AAC-hbr, sizelength=(string)13, indexlength=(string)3, indexdeltalength=(string)3, config=(string)121056E500"
SDP
function create_sdp {
local sdp_path=$1
# sdp
cat >$sdp_path <<EOF
v=0
c=IN IP4 ${dst_address}
m=video ${video_rtp_port} RTP/AVP ${rtp_video_payload_type}
a=rtpmap:${rtp_video_payload_type} ${video_media_subtype}/90000
m=audio ${audio_rtp_port} RTP/AVP ${rtp_audio_payload_type}
a=rtpmap:${rtp_audio_payload_type} ${audio_media_subtype}/${audio_rate}/${audio_channels}
EOF
if (( channels == 2 )) && [[ ${audio_media_subtype} == "opus" ]]
then
echo "a=fmtp:${rtp_audio_payload_type} sprop-stereo=1" >>${sdp_path}
fi
}
RTP
gst-launch-1.0 \
filesrc location=${input_path} ! qtdemux name=demux \
demux.video_0 ! queue ! rtph264pay pt=$rtp_video_payload_type ! \
udpsink host=${dst_address} port=${video_rtp_port} sync=true \
demux.audio_0 ! queue ! rtpmp4gpay pt=$rtp_audio_payload_type ! \
udpsink host=${dst_address} port=${audio_rtp_port} sync=true
gst-launch-1.0 \
udpsrc address=${address} port=${video_rtp_port} do-timestamp=true ! "$VIDEOCAPS" ! \
"application/x-rtp, media=(string)video"! queue ! decodebin ! autovideosink sync=true \
udpsrc address=${address} port=${audio_rtp_port} do-timestamp=true ! "$AUDIOCAPS" ! \
"application/x-rtp, media=(string)audio"! queue ! decodebin ! autoaudiosink sync=true
RTP using rtpbin
gst-launch-1.0 \
rtpbin name=bin \
filesrc location=${input_path} ! qtdemux name=demux \
demux.video_0 ! queue ! rtph264pay pt=$rtp_video_payload_type ! bin.send_rtp_sink_0 \
bin.send_rtp_src_0 ! udpsink host=${dst_address} port=${video_rtp_port} sync=true \
demux.audio_0 ! queue ! rtpmp4gpay pt=$rtp_audio_payload_type ! bin.send_rtp_sink_1\
bin.send_rtp_src_1 ! udpsink host=${dst_address} port=${audio_rtp_port} sync=true
gst-launch-1.0 \
rtpbin name=bin \
udpsrc address=${address} port=${video_rtp_port} do-timestamp=true ! "$VIDEOCAPS" ! bin.recv_rtp_sink_0 \
bin. !"application/x-rtp, media=(string)video"! queue ! decodebin ! autovideosink sync=true \
udpsrc address=${address} port=${audio_rtp_port} do-timestamp=true ! "$AUDIOCAPS" ! bin.recv_rtp_sink_1 \
bin. !"application/x-rtp, media=(string)audio"! queue ! decodebin ! autoaudiosink sync=true
RTP+RTCP using rtpbin
gst-launch-1.0 \
rtpbin name=bin \
filesrc location=${input_path} ! qtdemux name=demux \
demux.video_0 ! queue ! rtph264pay pt=$rtp_video_payload_type ! bin.send_rtp_sink_0 \
bin.send_rtp_src_0 ! udpsink host=${dst_address} port=${video_rtp_port} sync=true \
bin.send_rtcp_src_0 ! udpsink host=${dst_address} port=${video_rtcp_port} sync=false async=false \
demux.audio_0 ! queue ! rtpmp4gpay pt=$rtp_audio_payload_type ! bin.send_rtp_sink_1\
bin.send_rtp_src_1 ! udpsink host=${dst_address} port=${audio_rtp_port} sync=true \
bin.send_rtcp_src_1 ! udpsink host=${dst_address} port=${audio_rtcp_port} sync=false async=false \
gst-launch-1.0 \
rtpbin name=bin \
udpsrc address=${address} port=${video_rtp_port} do-timestamp=true ! "$VIDEOCAPS" ! bin.recv_rtp_sink_0 \
bin. !"application/x-rtp, media=(string)video"! queue ! decodebin ! autovideosink sync=true \
udpsrc address=${address} port=${video_rtcp_port} ! "application/x-rtcp" ! bin.recv_rtcp_sink_0 \
udpsrc address=${address} port=${audio_rtp_port} do-timestamp=true ! "$AUDIOCAPS" ! bin.recv_rtp_sink_1 \
bin. !"application/x-rtp, media=(string)audio"! queue ! decodebin ! autoaudiosink sync=true \
udpsrc address=${address} port=${audio_rtcp_port} ! "application/x-rtcp" ! bin.recv_rtcp_sink_1
RTP+RTCP using sdpdemux
gst-launch-1.0 -v \
filesrc location=$sdp_path do-timestamp=true ! sdpdemux latency=${sdpdemux_latency_ms} name=bin \
bin. ! "application/x-rtp, media=(string)video" ! queue ! decodebin ! videoconvert ! videoscale ! queue ! videorate ! autovideosink sync=true \
bin. ! "application/x-rtp, media=(string)audio" ! queue ! decodebin ! audioconvert ! audioresample ! queue ! audiorate ! autoaudiosink sync=true
- encode webcam, UDP stream:
-
gst-launch v4l2src ! video/x-raw-yuv,width=128,height=96,format='(fourcc)'UYVY ! ffmpegcolorspace ! ffenc_h263 ! video/x-h263 ! rtph263ppay pt=96 ! udpsink host=192.168.1.1 port=5000 sync=false
- test VP8 / Opus to RTP (no RTPC) (WebRTC and Janus)
-
gst-launch-1.0 \
audiotestsrc is-live=true wave=5 ! audioresample ! audioconvert ! audio/x-raw,channels=2,rate=16000 ! opusenc bitrate=20000 ! rtpopuspay pt=97 ! udpsink host=127.0.0.1 port=5002 \
videotestsrc ! video/x-raw,width=320,height=240,framerate=15/1 ! videoscale ! videorate ! videoconvert ! timeoverlay ! vp8enc ! rtpvp8pay pt=96 ! udpsink host=127.0.0.1 port=5004- sdp
-
v=0
c=IN IP4 127.0.0.1
m=video 5100 RTP/AVP 96
a=rtpmap:96 VP8/90000
m=audio 5102 RTP/AVP 97
a=rtpmap:97 opus/48000/2
a=fmtp:97 sprop-stereo=1
- test VP8 / Opus to RTP (with RTCP, using rtpbin):
-
sdp_path=/tmp/toto.sdp
dst_address=225.4.3.2
video_rtp_port=5100
video_rtcp_port=$(( video_rtp_port + 1 ))
video_media_subtype="VP8"
rtp_video_payload_type=96
audio_rtp_port=$(( video_rtp_port + 2 ))
audio_rtcp_port=$(( video_rtp_port + 3 ))
audio_media_subtype="opus"
rtp_audio_payload_type=$(( rtp_video_payload_type + 1 ))
rate=48000
channels=2
# sdp
cat >$sdp_path <<EOF
v=0
c=IN IP4 $dst_address
m=video $video_rtp_port RTP/AVP $rtp_video_payload_type
a=rtpmap:$rtp_video_payload_type ${video_media_subtype}/90000
m=audio $audio_rtp_port RTP/AVP $rtp_audio_payload_type
a=rtpmap:$rtp_audio_payload_type ${audio_media_subtype}/${rate}/${channels}
EOF
if (( channels == 2 )) && [[ ${audio_media_subtype} == "opus" ]]
then
echo "a=fmtp:${rtp_audio_payload_type} sprop-stereo=1" >>${sdp_path}
fi
gst-launch-1.0 -v \
rtpbin name=bin \
videotestsrc ! video/x-raw,width=320,height=240,framerate=25/1 ! videoscale ! videorate ! videoconvert ! timeoverlay ! vp8enc ! rtpvp8pay pt=$rtp_video_payload_type ! bin.send_rtp_sink_0 \
bin.send_rtp_src_0 ! udpsink host=${dst_address} port=${video_rtp_port} sync=true \
bin.send_rtcp_src_0 ! udpsink host=${dst_address} port=${video_rtcp_port} sync=false async=false \
audiotestsrc is-live=true wave=5 ! audioconvert ! audioresample ! audio/x-raw,channels=${channels},rate=${rate} ! opusenc bitrate=64000 ! rtpopuspay pt=$rtp_audio_payload_type ! bin.send_rtp_sink_1 \
bin.send_rtp_src_1 ! udpsink host=${dst_address} port=${audio_rtp_port} sync=true \
bin.send_rtcp_src_1 ! udpsink host=${dst_address} port=${audio_rtcp_port} sync=false async=false
- test H.264 to RTP (no RTCP)
-
gst-launch-1.0 -v videotestsrc ! videoconvert ! x264enc ! rtph264pay config-interval=10 pt=96 ! udpsink host=234.1.2.3 port=5004- player:
-
- toto.sdp
-
v=0
m=video 5004 RTP/AVP 96
c=IN IP4 234.1.2.3
a=rtpmap:96 H264/90000
a=fmtp:96 packetization-mode=1
ffplay -i toto.sdp
- from file (VP8, Opus) to RTP (no RTCP)
-
sdp_path=/tmp/toto.sdp
dst_address=225.4.3.2
video_rtp_port=5100
video_rtcp_port=$(( video_rtp_port + 1 ))
video_media_subtype="VP8"
rtp_video_payload_type=96
audio_rtp_port=$(( video_rtp_port + 2 ))
audio_rtcp_port=$(( video_rtp_port + 3 ))
audio_media_subtype="opus"
rtp_audio_payload_type=$(( rtp_video_payload_type + 1 ))
rate=48000
channels=2
# sdp
cat >$sdp_path <<EOF
v=0
c=IN IP4 $dst_address
m=video $video_rtp_port RTP/AVP $rtp_video_payload_type
a=rtpmap:$rtp_video_payload_type ${video_media_subtype}/90000
m=audio $audio_rtp_port RTP/AVP $rtp_audio_payload_type
a=rtpmap:$rtp_audio_payload_type ${audio_media_subtype}/${rate}/${channels}
EOF
if (( channels == 2 )) && [[ ${audio_media_subtype} == "opus" ]]
then
echo "a=fmtp:${rtp_audio_payload_type} sprop-stereo=1" >>${sdp_path}
fi
gst-launch-1.0 -v \
filesrc location=/path/to/toto.webm ! matroskademux name=demux \
demux.video_0 ! queue ! rtpvp8pay pt=$rtp_video_payload_type ! udpsink host=${dst_address} port=${video_rtp_port} sync=true \
demux.audio_0 ! queue ! rtpopuspay pt=$rtp_audio_payload_type ! udpsink host=${dst_address} port=${audio_rtp_port} sync=true
- from file (H.264, AAC) to RTP
-
gst-launch \
filesrc location=${input_path} ! qtdemux name=demux \
demux.video_0 ! queue ! rtph264pay pt=$rtp_video_payload_type ! udpsink host=${dst_address} port=${video_rtp_port} sync=true \
demux.audio_0 ! queue ! rtpmp4gpay pt=$rtp_audio_payload_type ! udpsink host=${dst_address} port=${audio_rtp_port} sync=true
- from file (H.264, AAC) to RTP, using rtpbin
-
- ...
- ...
- from file (H.264, AAC) to RTP + RTCP,
using rtpbin
-
- ...
- ...
- RTP
and RTSP support
- RTMP to:
-
- options from librtmp
- nginx-rtmp
-
- Video and audio:
-
- from test to H.264, AAC
-
gst-launch-1.0 -v flvmux name=mux ! rtmpsink location=rtmp://nginx-server/myapp/mystream \
videotestsrc ! video/x-raw, width=360, height=288 ! x264enc ! video/x-h264,profile=baseline,width=360,height=288 ! h264parse ! mux. \
audiotestsrc wave=5 ! audioconvert ! avenc_aac compliance=experimental ! aacparse ! mux.
- from test to H.264 (omx), MP3
-
gst-launch-1.0 -v flvmux name=mux ! rtmpsink location=rtmp://nginx-server/myapp/mystream \
videotestsrc ! video/x-raw, width=360, height=288 ! omxh264enc ! video/x-h264,profile=baseline,width=360,height=288 ! h264parse ! mux. \
audiotestsrc wave=5 ! audioconvert ! lamemp3enc ! mpegaudioparse ! mux.
- from DVB
-
gst-launch-1.0 -v dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux \
flvmux name=mux ! rtmpsink location=rtmp://nginx-server/myapp/mystream \
demux. ! queue ! mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=320, height=320 ! videoconvert ! omxh264enc inline-header=true periodicty-idr=50 ! h264parse ! mux. \
demux. ! queue ! mpegaudioparse ! decodebin ! audioconvert ! avenc_aac compliance=experimental ! aacparse ! mux.
- video, audio with PID 0x7c:
-
gst-launch-1.0 -vvv dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux \
flvmux name=mux ! rtmpsink location=rtmp://192.168.0.8/myapp/mystream \
demux. ! queue ! mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=320, height=320 ! videoconvert ! omxh264enc inline-header=true periodicty-idr=50 ! h264parse ! mux. \
demux.audio_007c ! queue ! mpegaudioparse ! decodebin ! audioconvert ! avenc_aac compliance=experimental ! aacparse ! mux.
- video, test audio:
-
gst-launch-1.0 -vvv dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux \
flvmux name=mux ! rtmpsink location=rtmp://nginx-server/myapp/mystream \
demux. ! queue ! mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=320, height=320 ! videoconvert ! omxh264enc inline-header=true periodicty-idr=50 ! h264parse ! mux. \
audiotestsrc wave=5 ! audioconvert ! lamemp3enc ! mpegaudioparse ! mux.
- video, audio forced to 44100Hz (MP3 at 48000Hz is not supported by FLV)
(AAC at 48000Hz is supported, though) (queue
max-size-time must be increase from 1000000ns [1s] to ... )
-
gst-launch-1.0 -vvv dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux \
flvmux name=mux ! rtmpsink location=rtmp://nginx-server/myapp/mystream \
demux. ! queue max-size-time=4000000000 ! mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=320, height=320 ! videoconvert ! omxh264enc inline-header=true periodicty-idr=50 ! h264parse ! mux. \
demux.audio_007c ! queue max-size-time=4000000000 ! mpegaudioparse ! decodebin ! audioconvert ! audioresample ! audio/x-raw,rate=44100 ! lamemp3enc ! mpegaudioparse ! mux.
- video, audio AAC at 48000Hz:
-
gst-launch-1.0 -vvv dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=806 name=demux \
flvmux name=mux ! rtmpsink location=rtmp://nginx-server/myapp/mystream \
demux. ! queue ! mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=32, height=32 ! videoconvert ! omxh264enc inline-header=true periodicty-idr=50 ! h264parse ! mux. \
demux.audio_007c ! queue ! mpegaudioparse ! decodebin ! audioconvert ! audioresample ! audio/x-raw,rate=48000 ! avenc_aac compliance=experimental ! aacparse ! mux.
- Problemes / Problems
-
- queues
-
- [gst-devel]
DV capture pipeline frozen
- Solution:
-
export GST_DEBUG="*queue*:5"
- disable maximum limits
in queue:
-
queue max-size-bytes=0 max-size-buffers=0 max-size-time=0
- [gst-devel]
DV capture pipeline frozen
- when sending to nginx-rtmp-module, with nginx logs at level info (
error_log /var/log/nginx/error.log info): RTMP in chunk stream too big: xx >= 32too big message ...- too big message: 1048576 #324
- Solució / Solution
- nginx.conf
rtmp {
server {
# compatibility with GStreamer:
publish_time_fix on;
...
}
}chunk stream too big-
- I got
"RTMP in chunk stream too big: 64 >= 32" error when send rtmp stream
to nginx #246
- Broadcasting a video playlist to nginx-rtmp-module
Solution: nginx.conf
-
rtmp {
server {
# compatibility with GStreamer and Android from Samsung S5 using ...:
publish_time_fix off;
# only for publish-time-fix2 branch:
#publish_time_fix2 on;
...
}
}/usr/local/nginx/sbin/nginx -s reload
- I got
"RTMP in chunk stream too big: 64 >= 32" error when send rtmp stream
to nginx #246
-
- Wowza
-
- Live Streaming from RaspberryPi using GStreamer - Help please?
-
- Incoming security / Flash Version String:
-
Wirecast/|FME/|FMLE/|Wowza GoCoder*|Gstreamer/|Gstreamer/*|Gstreamer*
- How to secure publishing from an RTMP encoder that does not support authentication (ModuleSecureURLParams)
- Streaming
to a Flash Media Server using the rtmpsink element
-
- profile=baseline
- profile=baseline
- woking / not working
-
- working
-
- omxh264enc ! video/x-h264,profile=high
- omxh264enc ! video/x-h264,profile=baseline
- x264enc ! video/x-h264,profile=baseline
- not working
-
- x264enc !
video/x-h264,profile=high
- x264enc !
video/x-h264,profile=high
- only video:
-
gst-launch-1.0 -v -e flvmux name=mux ! rtmpsink location=rtmp://wowza_server/application/stream \
videotestsrc ! video/x-raw, framerate=25/1, width=640, height=360 ! x264enc bitrate=512 ! video/x-h264,profile=baseline ! h264parse ! mux.gst-launch-1.0 -v dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=805 name=demux \
flvmux name=mux ! rtmpsink location=rtmp://wowza_server/application/stream \
demux. ! queue ! mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=360, height=288 ! videoconvert ! omxh264enc inline-header=true periodicty-idr=1 ! video/x-h264,profile=baseline ! h264parse ! mux.gst-launch-1.0 -v dvbsrc modulation="QAM 64" trans-mode=8k bandwidth=8 frequency=658000000 code-rate-lp=AUTO code-rate-hp=2/3 guard=4 hierarchy=0 ! tsdemux program-number=805 name=demux \
flvmux name=mux ! rtmpsink location=rtmp://wowza-server/application/stream \
demux. ! queue ! mpegvideoparse ! decodebin ! videoscale ! video/x-raw, width=360, height=288 ! videoconvert ! omxh264enc inline-header=true periodicty-idr=1 ! video/x-h264,profile=high ! h264parse ! mux.gst-launch-1.0 -v flvmux name=mux ! rtmpsink location=rtmp://wowza-server/application/stream \
videotestsrc ! video/x-raw, width=360, height=288 ! omxh264enc ! video/x-h264,profile=high ! h264parse ! mux.
- video and audio:
-
gst-launch-1.0 -v -e flvmux name=mux ! rtmpsink location=rtmp://wowza-server/application/stream videotestsrc ! video/x-raw, framerate=24/1, width=1024, height=436 ! x264enc bitrate=800 ! video/x-h264,profile=baseline ! h264parse ! mux. audiotestsrc wave=5 ! audioconvert ! lamemp3enc ! mpegaudioparse ! mux.gst-launch-1.0 -v flvmux name=mux ! rtmpsink location=rtmp://wowza-server/application/stream videotestsrc ! video/x-raw, width=360, height=288 ! x264enc ! video/x-h264,profile=baseline ! h264parse ! mux. audiotestsrc wave=5 ! audioconvert ! lamemp3enc ! mpegaudioparse ! mux.gst-launch-1.0 -v flvmux name=mux ! rtmpsink location=rtmp://wowza-server/application/stream videotestsrc ! video/x-raw, width=360, height=288 ! omxh264enc ! video/x-h264,profile=baseline ! h264parse ! mux. audiotestsrc wave=5 ! audioconvert ! lamemp3enc ! mpegaudioparse ! mux.gst-launch-1.0 -v flvmux name=mux ! rtmpsink location=rtmp://wowza-server/application/stream videotestsrc ! video/x-raw, width=360, height=288 ! omxh264enc ! video/x-h264,profile=baseline ! h264parse ! mux.
- Flash Media Server
-
- Streaming to a Flash Media Server using the rtmpsink element
-
- audio/mpeg,mpegversion=4,stream-format=raw
- options from librtmp
- Introduction
to network streaming using GStreamer
gst-launch filesrc location=videofile ! decodebin name=decoder \
decoder. ! queue ! audioconvert ! audioresample ! osssink \
decoder. ! ffmpegcolorspace ! xvimagesink
- capture timestamped frames (BeagleCam)
-
gst-launch v4l2src num-buffers=1 ! video/x-raw-yuv,width=640,height=480,framerate=30/1 ! ffmpegcolorspace ! jpegenc ! filesink location=$(date +"%s").jpg
- gstreamer dvb streaming
- Graphical editor
- gst-editor (only for gstreamer 0.8)
Desenvolupament / Development
- Documentació / Documentation
- Llenguatges / Languages
- C applications
(using GObject
and GLib)
-
- Tutorials
-
- Building applications using GStreamer
-
gcc basic-tutorial-1.c -o basic-tutorial-1 `pkg-config --cflags --libs gstreamer-1.0`
- Basic tutorials
-
- ...
- Basic tutorial 11: Debugging tools
- Basic
tutorial 12: Streaming (same as basic-tutorial-1, but with buffer
watch)
- ...
- Basic tutorial 14: Handy elements
-
- audiorate: it takes an incoming stream of time-stamped raw audio frames and produces a perfect stream by inserting or dropping samples as needed
- API
-
- GstMessage
- Python applications (using PyGObject)
-
- Configuració / Setup
- Python with PyGObject
- GStreamer
- from packages
- ...
- or compiled and installed to /usr/local/bin
- check that you compiled GStreamer with
PyGObject enabled
- you have to export:
export GI_TYPELIB_PATH=/usr/local/lib/girepository-1.0- PyGObject API
- Tutorials and examples
-
- Basic tutorials
- gkralik / python-gst-tutorial
-
- basic-tutorial-...py
- basic-tutorial-...py
- rubenrua / GstreamerCodeSnippets / Python /
-
- pygst-sdk-tutorials
-
- basic-tutorial-...py
- basic-tutorial-...py
- pygst-tutorial
- How te retrieve stream statistics in Gstreamer?
-
- RTP/RTCP server and client
gst_bus_add_signal_watch
- Basic tutorials
- Problemes / Problems
-
- CTRL-C not working
- Solució / Solution
if __name__ == '__main__':
# to be able to use CTRL-C to quit
import signal
signal.signal(signal.SIGINT, signal.SIG_DFL)
...
ValueError: Namespace Gst not available-
- Solució / Solution
-
- if you compiled gstreamer from
source,
-
- check that typelib files where
installed (
lib64girepository-develmust be installed before compilation):
-
- /usr/local/lib/girepository-1.0/Gst*.typelib
- set environment variable:
-
export GI_TYPELIB_PATH=/usr/local/lib/girepository-1.0
- check that typelib files where
installed (
- if you compiled gstreamer from
source,
- aaaa
- aaaa
- GES
- CLI: ges-launch-1.0
- PyGObject API
- Exemples / Examples
- ...
- ...
- Resum / Summary
task
options
steps
C
Python
methods
headers
#include <gst/gst.h>
#!/usr/bin/env python3
import sys
import gi
gi.require_version('Gst', '1.0')
from gi.repository import Gst, GObject, GLib
Initialize Gstreamer
/* init */
gst_init (&argc, &argv);
class Player(object):def __init__(self):
# init GStreamer
Gst.init(None)
main GLib loop
(optional, bu t needed when using add_signal_watch)
GMainLoop *main_loop;
main_loop = g_main_loop_new (NULL, FALSE);
self.loop = GLib.MainLoop.new(None, False)Arguments
Usage /* check args */
if (argc != 2) {
g_print ("Usage: %s <filename>\n", argv[0]);
return -1;
}
# check input arguments
if len(sys.argv) != 2:
print("Usage: {0:s} <filename>".format(sys.argv[0]))
sys.exit(1)
Arguments
if Gst.Uri.is_valid(sys.argv[1]):
uri = sys.argv[1]
else:
uri = Gst.filename_to_uri(sys.argv[1])
print("uri: {0:s}".format(uri))Build the pipeline
option 1:
Build the pipeline by parsing (basic tutorial 1)
/* create a new pipeline with all the elements */
GstElement *pipeline;
pipeline = gst_parse_launch ("playbin uri=https://www.freedesktop.org/software/gstreamer-sdk/data/media/sintel_trailer-480p.webm")
# create a new pipeline with all the elements
pipeline = Gst.parse_launch("playbin uri=https://www.freedesktop.org/software/gstreamer-sdk/data/media/sintel_trailer-480p.webm")
option 2:
Build the pipeline from elements
(basic tutorial 2)
Create the elements /* create the elements */
GstElement *pipeline, *source, *sink;
source = gst_element_factory_make ("videotestsrc", "source");
sink = gst_element_factory_make ("autovideosink", "sink");
# create the elements
source = Gst.ElementFactory.make("videotestsrc", "source")
sink = Gst.ElementFactory.make("autovideosink", "sink")
(or create the factory and then the element)
(Creating a GstElement, basic tutorial 6)
/* create factory and element */
factory = gst_element_factory_find ("fakesrc");
element = gst_element_factory_create (factory, "source");
# create factory and element
source_factory = Gst.ElementFactory.find("videotestsrc")
sink_factory = Gst.ElementFactory.find("autovideosink")
source = source_factory.create("source")
sink = sink_factory.create("sink")
Create the empty pipeline /* create an empty pipeline */
pipeline = gst_pipeline_new ("test-pipeline");
# create an empty pipeline
pipeline = Gst.Pipeline.new("test-pipeline")
Build the pipeline: add and link elements
/* add elements to the pipeline */
gst_bin_add_many (GST_BIN (pipeline), source, sink, NULL);
if (gst_element_link (source, sink) != TRUE) {
g_printerr ("Elements could not be linked.\n");
gst_object_unref (pipeline);
return -1;
}
# add elements to the pipeline
pipeline.add(source)
pipeline.add(sink)
if not source.link(sink):
print("ERROR: Could not link source to sink")
sys.exit(1)
Modify properties /* set property of an element */
g_object_set (source, "pattern", 0, NULL);
# set property of an element
source.set_property("pattern", 0)
Connect element signal to a callback
def on_have_type(self, element, probability, caps, user_data):
# connect signal to a callback
typefind.connect("have-type", on_have_type, None)
option 3:
Dinamically connect the elements in the pipeline
(basic tutorial 3)
Callback pad_added_handler(...)
def on_pad_added(self, ...):
Connect signal to callback g_signal_connect (data.source, "pad-added", G_CALLBACK (pad_added_handler), &data);
# connect signal to a callback
source.connect("pad-added", self.on_pad_added)
Start playing
/* set the pipeline to playing state */
gst_element_set_state (pipeline, GST_STATE_PLAYING);
# set the pipeline to playing state
pipeline.set_state(Gst.State.PLAYING)
Main loop
(two ways to use a bus)
Get bus associated to pipeline /* get the bus from the pipeline */
GstBus *bus;
bus = gst_element_get_bus (pipeline);# get the bus from the pipeline
bus = pipeline.get_bus()option 1:
Wait until error or EOS
(basic tutorial 1)
/* wait until error or end of stream */
GstMessage *msg;
msg = gst_bus_timed_pop_filtered (bus, GST_CLOCK_TIME_NONE, GST_MESSAGE_ERROR | GST_MESSAGE_EOS);
# wait until error or end of stream
terminate = False
while True:
try:
msg = bus.timed_pop_filtered(0.5 * Gst.SECOND, Gst.MessageType.ERROR | Gst.MessageType.EOS)
if msg:
terminate = True
except KeyboardInterrupt:
terminate = True
if terminate:
break
option 2:
GMain loop with callback
(basic tutorial 12)
start loop /* start main loop */
g_main_loop_run (main_loop);# start main loop
self.loop.run()data structure (to be available from callback)
typedef struct _CustomData {
gboolean is_live;
GstElement *pipeline;
GMainLoop *loop;
} CustomData;
CustomData data;
data.loop = main_loop;
data.pipeline = pipeline;
self.loop
self.pipeline
option a: one single callback for all messages
static void cb_message (GstBus *bus, GstMessage *msg, CustomData *data) {
const GstStructure *structure;
structure = gst_message_get_structure (msg);
g_print ("Message name: %s\n", gst_structure_get_name (structure) );
switch (GST_MESSAGE_TYPE (msg)) {
...
}def on_message(self, bus, msg, user_data):
t = msg.type
if (t==Gst.MessageType.EOS):
print("End of stream")
self.loop.quit()
elif (t == Gst.MessageType.ERROR):
err, debug = msg.parse_error()
print("Error: {0:s}".format(err.message))
self.loop.quit()
else:
# Unhandled message
pass
/* message handler */
gst_bus_add_signal_watch (bus);
g_signal_connect (bus, "message", G_CALLBACK (cb_message), &data);
# general message handler
bus.add_watch(GLib.PRIORITY_DEFAULT, self.on_message, None)
option b: each message has its callback
def on_error(self,bus, msg, user_data):
def on_eos(self,bus, msg, user_data):
def on_state_changed(self,bus, msg, user_data):
def on_application_message(self,bus, msg, user_data):
...
# individual message handler
bus.add_signal_watch()
bus.connect("message::error", self.on_error)
bus.connect("message::eos", self.on_eos)
bus.connect("message::state-changed", self.on_state_changed)
bus.connect("message::application", self.on_application_message)
...
quit loop
self.loop.quit()Free resources
/* free resources */
if (msg != NULL)
gst_message_unref (msg);
gst_object_unref (bus);
gst_element_set_state (pipeline, GST_STATE_NULL);
gst_object_unref (pipeline);
return 0;
# free resources
pipeline.set_state(Gst.State.NULL)
if __name__ == '__main__':
# to be able to use CTRL-C to quit
import signal
signal.signal(signal.SIGINT, signal.SIG_DFL)
p = Player()
- Application Development Manual
- ------ inici destinació ------
- Application Development Manual (pdf, ps, html)
- About GStreamer
- What is GStreamer?
- Design principles
- Foundations
- Communication between application / bus / pipeline:
- buffers: streaming data between
elements (downstream (sources->sinks)) (buffering)
- events: between elements or from the
application to elements (upstream (sinks->sources), downstream
(sources->sinks))
- messages: posted by elements on the
pipeline's message bus (message types)
- queries:
allow applications to request information such as duration or current
playback position from the pipeline (upstream, downstream) (querying)
- Building an Application
- Initializing GStreamer
-
C
Python #include <gst/gst.h>
import gi
gi.require_version('Gst', '1.0')
from gi.repository import Gst, GObject, GLib
gst_init (&argc, &argv);
Gst.init(None)
- Elements
- Creating elements
C
Python option 1: factory and element
factory = gst_element_factory_find("fakesrc")
element = gst_element_factory_create(factory, "source")
factory = Gst.ElementFactory.find("fakesrc")
source = factory.create("source")
option 2: element
element = gst_element_factory_make("fakesrc", "source")
source = Gst.ElementFactory.make("fakesrc", "source")
- Element States
name
C
Python description
gst_element_set_state (pipeline, GST_STATE_PLAYING);pipeline.set_state(Gst.State.PLAYING)
NULL
GST_STATE_NULL
Gst.State.NULL
the NULL state or initial state of an element. Transition to it will free all resources.
READY
GST_STATE_READYGst.State.READYthe element is ready to go to PAUSED
PAUSED
GST_STATE_PAUSEDGst.State.PAUSEDthe element is PAUSED, it is ready to accept and process data. Sink elements however only accept one buffer and then block
PLAYING
GST_STATE_PLAYINGGst.State.PLAYINGthe element is PLAYING, the clock is running and the data is flowing
- Bins
-
C
Python
Bin
gst_bin_new()Gst.Bin.new()gst_bin_add()
gst_bin_remove()
gst_bin_get_by_name()
gst_bin_get_by_interface()
gst_bin_iterate_elements()
Pipeline
(special top-level type of bin)
gst_pipeline_new()pipeline = Gst.Pipeline.new("test-pipeline")
pipeline.add(...)
...
- Bus
-
C
Python get all messages
bus = gst_pipeline_get_bus (GST_PIPELINE (pipeline));
bus_watch_id = gst_bus_add_watch (bus, my_bus_callback, NULL);
static gboolean my_bus_callback(GstBus *bus, GstMessage *msg, gpointer data)
...
bus = self.pipeline.get_bus()bus.add_watch(GLib.PRIORITY_DEFAULT, self.on_message, None)
def on_message(bus, msg, user_data):
if (msg.type == Gst.MessageType.ERROR):
err, debug = msg.parse_error()
print("Error: {0:s}".format(err.message))
self.loop.quit()
elif (msg.type == Gst.MessageType.EOS):
print("EOS")
self.loop.quit()
else:
# Unhandled message
pass
get selected messages
bus = gst_pipeline_get_bus (GST_PIPELINE (pipeline);
gst_bus_add_signal_watch (bus);
g_signal_connect (bus, "message::error", G_CALLBACK (cb_message_error), NULL);
g_signal_connect (bus, "message::eos", G_CALLBACK (cb_message_eos), NULL);
bus = self.pipeline.get_bus()
bus.add_signal_watch()
bus.connect("message::error", self.on_error)
bus.connect("message::eos", self.on_eos)
(needed loop)
loop = g_main_loop_new (NULL, FALSE);
g_main_loop_run (loop);
...
g_main_loop_unref (loop);
loop = GLib.MainLoop.new(None, False)
loop.run()
...
loop.quit()
- Tipus de missatges
/ Message
types
-
message type
connect(...)
C
Python
Gst.MessageType
Error (fatal problem),
warning (non-fatal problem),
information (not problem)
message::error
GST_MESSAGE_ERROR
gst_message_parse_error()
_parse_warning ()
_parse_info ()
Gst.MessageType.ERROR
parse_error()
parse_warning()
parse_info()
End-of-stream
message::eos
GST_MESSAGE_EOS
Gst.MessageType.EOSTags
GST_MESSAGE_TAG
gst_message_parse_tag()
Gst.MessageType.TAG
tags = msg.parse_tag()
tags.foreach(self.print_one_tag, None)
def print_one_tag(self, list, tag, user_data):
res, val = Gst.TagList.copy_value(list, tag)
print("%s: %s"% (Gst.tag_get_nick(tag), val))
State-changes
GST_MESSAGE_STATE_CHANGE
gst_message_parse_state_changed ()
Gst.MessageType.STATE_CHANGED
old, new, pending = msg.parse_state_changed()
print("State changed: {0:s} -> {1:s}".format(Gst.Element.state_get_name(old), Gst.Element.state_get_name(new)))
message::async-done
GST_MESSAGE_ASYNC_DONEGst.MessageType.ASYNC_DONEBuffering
message::buffering
GST_MESSAGE_BUFFERING
gst_message_parse_buffering (message, &percent);
Gst.MessageType.BUFFERING
percent = message.parse_buffering()
Element
Application-specific
gst_message_get_structure()
Threads
sync-message::stream-status
GST_MESSAGE_STREAM_STATUS
gst_message_parse_stream_status (message, &type, &owner);
Gst.MessageType.STREAM_STATUS
type, owner = message.parse_stream_status()
- Pads and capabilities
- Pads
availability
examples
C
Python
pad = gst_element_get_static_pad(...)
pad = gst_element_get_compatible_pad (mux, tolink_pad, NULL);
pad = gst_element_get_request_pad (tee, "src%d");name = gst_pad_get_name (pad);
pad = convert.get_static_pad("sink")
pad = mux.get_compatible_pad(tolink_pad)pad = tee.get_request_pad("src%d")name = pad.get_name()
always
dynamic (sometimes) pads
- demuxer
/* listen for newly created pads */
g_signal_connect (demux, "pad-added", G_CALLBACK (cb_new_pad), NULL);static void cb_new_pad(...)gst_element_set_state ()gst_element_sync_state_with_parent ()
# listen for newly created pads
self.demux.connect("pad-added", self.on_new_pad)def on_new_pad(self):...
request pads
(basic tutorial 3)
- multiplexer
- aggregator
- tee
- Capabilities
of a pad
- GstCaps
- non-negotiated pad: one or more GstStructure
- negotiated pad: only one GstStructure (with fixed values)
- ...
- possible caps: obtained with gst-inspect
- allowed caps: subset of possible capabilities,
depending on the possible caps of the peer pad
- negotiated caps:
- Types of caps:
-
type
Gst.Structure
values
empty
0
ANY
simple
1
variable field types
fixed
1
no variable field types
-
C
Python check type
gst_caps_is_fixed (caps)
caps.is_fixed()
get structure
str = gst_caps_get_structure (caps, 0);
str = caps.get_structure(0)
get value
gst_structure_get_int (str, "width", &width)
width = str.get_int("width")
creation of simple caps
caps = gst_caps_new_simple ("video/x-raw",
"format", G_TYPE_STRING, "I420",
"width", G_TYPE_INT, 384,
"height", G_TYPE_INT, 288,
"framerate", GST_TYPE_FRACTION, 25, 1,
NULL);caps = Gst.Caps.new_empty_simple("video/x-raw")
caps.set_value("format", "I420")
caps.set_value("width", 384)
caps.set_value("height", 288)
caps.set_value("framerate", ...)
creation of full caps
caps = gst_caps_new_full (
gst_structure_new ("video/x-raw",
"width", G_TYPE_INT, 384,
"height", G_TYPE_INT, 288,
"framerate", GST_TYPE_FRACTION, 25, 1,
NULL),
gst_structure_new ("video/x-bayer",
"width", G_TYPE_INT, 384,
"height", G_TYPE_INT, 288,
"framerate", GST_TYPE_FRACTION, 25, 1,
NULL),
NULL);
(unavailable)
filtering using caps (internally creates a capsfilter)
link_ok = gst_element_link_filtered (element1, element2, caps);
link_ok = element1.link_filtered(element2, caps)
- Ghost pads
- "A ghost pad is a pad from some element in the bin that can be accessed directly from the bin as well."
-
C
Python gst_element_add_pad (bin, gst_ghost_pad_new ("sink", pad));
ghost_pad = Gst.GhostPad.new("sink",pad)
- Buffers and events
- Buffers
- Events
- "Events are control particles that are sent both upstream (right to left) and downstream (left to right) in a pipeline along with buffers."
- Examples: seeking, flushes, end-of-stream notifications, ...
-
C
Python create
event = gst_event_new_seek (1.0, GST_FORMAT_TIME,
GST_SEEK_FLAG_NONE,
GST_SEEK_METHOD_SET, time_ns,
GST_SEEK_TYPE_NONE, G_GUINT64_CONSTANT (0));
event = Gst.Event.new_seek(1.0, Gst.Format.TIME,
Gst.SeekFlags.NONE,
Gst.SeekType.SET, time_ns,
Gst.SeekType.NONE, 0)
send
gst_element_send_event (element, event);
element.send_event(event)
- Your first application
- Advanced GStreamer Concepts
- Position
tracking and seeking (basic_tutorial_4)
-
C
Python Querying (queries)
static gboolean cb_print_position(GstElement *pipeline)gst_element_query_position (pipeline, GST_FORMAT_TIME, &pos)gst_element_query_duration (pipeline, GST_FORMAT_TIME, &len)g_timeout_add (200, (GSourceFunc) cb_print_position, pipeline);
g_timeout_add_seconds (1, (GSourceFunc) cb_print_position, pipeline);
g_main_loop_run (loop);
def cb_print_position(pipeline)ret, pos = pipeline.query_position(Gst.Format.TIME)ret, len = pipeline.query_duration(Gst.Format.TIME)print("{0:f}/{1:f}".format(pos, len))
GLib.timeout_add(200, cb_print_position, pipeline)- the function is called repeatedly until
it returns False
GLib.timeout_add_seconds(1, cb_print_position, pipeline)loop.run()
Events: seeking (and more)
gst_element_seek (pipeline, 1.0,
GST_FORMAT_TIME,
GST_SEEK_FLAG_FLUSH,
GST_SEEK_TYPE_SET, time_nanoseconds,
GST_SEEK_TYPE_NONE, GST_CLOCK_TIME_NONE)gst_element_seek_simple (...)
pipeline.seek(1.0,
Gst.Format.TIME,
Gst.SeekFlags.FLUSH,
Gst.SeekType.SET, time_nanoseconds,
Gst.SeekType.NONE, Gst.CLOCK_TIME_NONE)
pipeline.seek_simple(
Gst.Format.TIME,
Gst.SeekFlags.FLUSH | Gst.SeekFlags.KEY_UNIT,
30 * Gst.SECOND)
- Metadata
- Types
- stream tags: non-technical information (author,
title, album ...)
- stream-info: technical information (GstPad, GstCaps)
-
C
Python Metadata reading
tags.py
Tag writing
- Interfaces
- GstColorBalance
- GstVideoOverlay
- ...
- Clocks and synchronization in GStreamer
- Buffering
-
Stream buffering "Buffering up to a specific amount of data, in memory, before starting playback so that network fluctuations are minimized" buffer element: queue2
- low watermark
- high watermark
Download buffering "Download of the network file to a local disk with fast seeking in the downloaded data. This is similar to the quicktime/youtube players." buffering.py Timeshift buffering "Caching of (semi)-live streams to a local, on disk, ringbuffer with seeking in the cached area. This is similar to tivo-like timeshifting."
-
C
Python
buffering.py (download buffering)
messages
gst_message_parse_buffering (message, &percent);
percent = message.parse_buffering()
queries
query = gst_query_new_buffering (GST_FORMAT_TIME);
gst_element_query (pipeline, query)
gst_query_parse_buffering_percent (query, &busy, &percent);
gst_query_parse_buffering_range (query, NULL, NULL, NULL, &estimated_total);
query = Gst.Query.new_buffering(Gst.Format.TIME)
pipeline.query(query)
busy, percent = query.parse_buffering_percent()
format, start, stop, estimated_total = query.parse_buffering_range()
- Dynamic Controllable Parameters
-
C
Python
GstControlSource
Gst.ControlSource
create
csource = gst_interpolation_control_source_new ();
g_object_set (csource, "mode", GST_INTERPOLATION_MODE_LINEAR, NULL);
attach to the gobject property
gst_object_add_control_binding (object, gst_direct_control_binding_new (object, "prop1", csource));
GstTimedValueControlSource *tv_csource = (GstTimedValueControlSource *)csource;
gst_timed_value_control_source_set (tv_csource, 0 * GST_SECOND, 0.0);
gst_timed_value_control_source_set (tv_csource, 1 * GST_SECOND, 1.0);
- Threads
- Scheduling in GStreamer
- pad can:
- push from upstream
- pull to downstream
- Configuring threads in GStreamer
- message: STREAM_STATUS
- GST_STREAM_STATUS_TYPE_CREATE: when a new thread
is about to be created -> you can configure a GstTaskPool in the
GstTask
- when a thread is entered or left -> you can configure thread priorities
- when a thread starts, pauses and stops -> you can visualize the status of streaming in a gui application
- Boost priority of a thread
- When would you force a thread?
- queue2 (Stream
buffering)
- queue
- Autoplugging
- Playback components
- Media types as a way to identify streams
- capability:
media types + set of properties
- Full list of Media
Types and Properties
- Media stream type detection
- Dynamically autoplugging a pipeline
- Pipeline manipulation
- Using probes
- Data probes
- Play a section of a media file
- Manually adding or removing data from/to a pipeline
- Forcing a format
- Dynamically changing the pipeline
- Higher-level interfaces for GStreamer applications
- Appendices
- ...
- Tutorials (source code in gst-docs/examples/tutorials)
(playback tutorials are based on playbin)
...
C
Python Table of Concepts
Basic tutorials / Playback tutorials
francesc.pinyol.m/python-gst-examples
gkralik / python-gst-tutorial
GstreamerCodeSnippets
Python/pygst-sdk-tutorialsGstreamerCodeSnippets
Others/0.10/Python/pygst-sdk-tutorialsGstreamerCodeSnippets
pygst-tutorial
(class GTK_Main)
Basic tutorial 1: Hello world!
basic-tutorial-1.py basic-tutorial-1.py
Bus
Elements
Links
Pipelines
Basic tutorial 2: GStreamer concepts
basic-tutorial-2.py
basic-tutorial-2-ex-vertigo.py
basic-tutorial-2.py
Pads
Signals
States
Basic tutorial 3: Dynamic pipelines
- CustomData data
- data.source
- data.convert
- data.sink
- g_signal_connect (data.source, "pad-added",
G_CALLBACK (pad_added_handler), &data);
basic-tutorial-3-ex-video.py
basic-tutorial-3.py
- class Player
- def __init__(self)
- self.source
- self.convert
- self.sink
- self.source.connect("pad-added",
self.on_pad_added)
- def on_pad_added(self, src, new_pad)
basic-tutorial-3.py
- def pad_added_handler(src, new_pad, data)
- data["source"]
- data["convert"]
- data["sink"]
- data["source"].connect("pad-added", pad_added_handler, data)
(you can jump to Playback tutorials)
Queries
Seeks
Basic tutorial 4: Time management
basic-tutorial-4.py
- class Player
- def __init__
- def play
- bus = self.playbin.get_bus()
- msg = bus.timed_pop_filtered(100 *
Gst.MSECOND, (Gst.MessageType.STATE_CHANGED | Gst.MessageType.ERROR |
Gst.MessageType.EOS | Gst.MessageType.DURATION_CHANGED))
- if msg: self.handle_message(msg)
- self.playbin.query_position
- self.playbin.query_duration
- self.playbin.seek_simple
- def handle_message
- query = Gst.Query.new_seeking(Gst.Format.TIME)
- if self.playbin.query(query):
basic-tutorial-4.py
GUI
Basic tutorial 5: GUI toolkit integration
- gst_bus_add_signal_watch (bus);
basic-tutorial-5.py
- class Player
- def __init__
- Gtk.init(sys.argv)
- Gst.init(sys.argv)
- # connect to interesting signals in playbin
self.playbin.connect("video-tags-changed", self.on_tags_changed)
self.playbin.connect("audio-tags-changed", self.on_tags_changed)
self.playbin.connect("text-tags-changed", self.on_tags_changed)
- # instruct the bus to emit signals for each
received message
# and connect to the interesting signals
bus = self.playbin.get_bus()
bus.add_signal_watch()
bus.connect("message::error", self.on_error)
bus.connect("message::eos", self.on_eos)
bus.connect("message::state-changed", self.on_state_changed)
bus.connect("message::application", self.on_application_message) - def start
- GLib.timeout_add_seconds(1, self.refresh_ui)
- Gtk.main()
- def cleanup
- def build_ui
- def on_realize
- def on_play
- def on_pause
- def on_stop
- ...
- def on_tags_changed
- self.playbin.post_message
- def on_error
- def on_eos
- def on_state_changed
- ...
- def analyze_streams
Capabilities
Basic tutorial 6: Media formats and Pad Capabilities
- static void print_pad_capabilities
- /* Retrieve negotiated caps (or acceptable
caps if negotiation is not finished yet) */
caps = gst_pad_get_current_caps (pad);
if (!caps)
caps = gst_pad_query_caps (pad, NULL);
basic-tutorial-6.py
- def print_field
- def print caps(caps, pfx)
- structure = caps.get_structure(i)
- structure.foreach(print_field, pfx)
- def print_pad_templates_information(factory)
- pads = factory.get_static_pad_templates()
- for pad in pads:
- padtemplate = pad.get()
- if padtemplate.get_caps():
- print_caps(padtemplate.get_caps(),
" ")
- def print_pad_capabilities(element, pad_name)
- pad = element.get_static_pad(pad_name)
- # retrieve negotiated caps (or acceptable
caps if negotiation is not yet finished)
caps = pad.get_current_caps()
if not caps:
caps = pad.get_allowed_caps()
- print_caps(caps,
" ")
- def main
- sink_factory = Gst.ElementFactory.find("autoaudiosink")
- print_pad_templates_information(sink_factory)
- sink = sink_factory.create("sink")
- print_pad_capabilities(sink, "sink")
Pad availability
- always
- sometimes
- on request
Basic tutorial 7: Multithreading and Pad Availability
basic-tutorial-7.py
- def main
- # manually link the tee, which has "Request"
pads
tee_src_pad_template = tee.get_pad_template("src_%u")
tee_audio_pad = tee.request_pad(tee_src_pad_template, None, None)
audio_queue_pad = audio_queue.get_static_pad("sink")
tee_audio_pad.link(audio_queue_pad)
tee_video_pad = tee.request_pad(tee_src_pad_template, None, None)
video_queue_pad = video_queue.get_static_pad("sink")
tee_video_pad.link(video_queue_pad)
Buffers
- GstBuffer: chunk of data. Can contain multiple
GstMemory (memory buffer)
Basic tutorial 8: Short-cutting the pipeline (same as 7, replacing audiotestsrc -> appsrc; adding a third branch appsink)
- appsrc
- appsink
- /* Configure appsrc */
gst_audio_info_set_format (&info, GST_AUDIO_FORMAT_S16, SAMPLE_RATE, 1, NULL);
audio_caps = gst_audio_info_to_caps (&info);
g_object_set (data.app_source, "caps", audio_caps, NULL);
g_signal_connect (data.app_source, "need-data", G_CALLBACK (start_feed), &data);
g_signal_connect (data.app_source, "enough-data", G_CALLBACK (stop_feed), &data);
Discoverer
Basic tutorial 9: Media information gathering
- GstDiscoverer
basic-tutorial-9.py
- class Discoverer
- def print_tag_foreach
- def print_stream_info
- def print_topology
- def on_discovered
- tags = info.get_tags()
- def on_finished
- self.loop.quit()
- def __init__
- self.loop = GLib.MainLoop.new(None, False)
- self.loop.run()
- get video size:
gst-discoverer-1.0
gst-launch-1.0
Tools
Basic tutorial 10: GStreamer tools
- gst-launch-1.0
- elements
- properties
- named elements
- pads
- caps filters
- gst-inspect-1.0
- gst-discoverer-1.0
Debugging
Basic tutorial 11: Debugging tools
Basic tutorial 12: Streaming
- Setting live streams to PAUSED succeeds, but
returns GST_STATE_CHANGE_NO_PREROLL, instead of
GST_STATE_CHANGE_SUCCESS to indicate that this is a live stream.
basic-tutorial-12.py
- class Player
- def on_message
- if (msg.type == Gst.MessageType.ERROR):
- elif (msg.type == Gst.MessageType.EOS):
- elif (msg.type == Gst.MessageType.BUFFERING):
- elif (msg.type == Gst.MessageType.CLOCK_LOST):
- def __init__
- self.pipeline = Gst.parse_launch(...)
- bus = self.pipeline.get_bus()
- ret = self.pipeline.set_state(Gst.State.PLAYING)
print("ret: {0:d}".format(ret))
if ret == Gst.StateChangeReturn.FAILURE:
print("Unable to set the pipeline to the playing state")
sys.exit(1)
elif ret == Gst.StateChangeReturn.NO_PREROLL:
self.is_live = True
- self.loop = GLib.MainLoop.new(None, False)
- bus.add_signal_watch()
bus.connect("message", self.on_message) - self.loop.run()
Basic tutorial 13: Playback speed
Basic tutorial 14: Handy elements
- Bins
- playbin
- uridecodebin
- decodebin
- File input/output
- filesrc
- filesink
- Network
- souphttpsrc
- Test media generation
- videotestsrc
- audiotestsrc
- Video adapters
- videoconvert
- videorate
- videoscale
- Audio adapters
- audioconvert
- audioresample
- audiorate
- Multithreading
- queue
- queue2
- multiqueue
- tee
- Capabilities
- Debugging
- fakesink
- identity
Basic tutorial 16: Platform-specific elements
Action signals
Audio switching
Tags
Playback tutorial 1: Playbin usage
- gst_bus_add_watch(bus);
Subtitles
Playback tutorial 2: Subtitle management
Playback tutorial 3: Short-cutting the pipeline
Playback tutorial 4: Progressive streaming
Playback tutorial 5: Color Balance
Playback tutorial 6: Audio visualization
Playback tutorial 7: Custom playbin sinks
Playback tutorial 8: Hardware-accelerated video decoding
Playback tutorial 9: Digital audio pass-through
- Encoding profiles and targets
- Exemples / Examples
- francesc.pinyol.m/python-gst-examples/
- ...
- Estructura / Structure
- ...
-
syntax
C
Python
Encoding target
- location of target files (*.gep)
$GST_DATADIR/gstreamer-GST_API_VERSION/encoding-profiles//usr/share/gstreamer-1.0/encoding-profiles/*.gep$HOME/gstreamer-GST_API_VERSION/encoding-profiles/~/.local/share/gstreamer-1.0/encoding-profiles/<category>/<name>.gep$GST_ENCODING_TARGET_PATH/- Pitivi:
/usr/share/pitivi/gstpresets/*.gep
- $(target.category)/$(target.name).gep
[GStreamer Encoding Target]
name : <encoding_target_name>
category : <category>
\description : <description> #translatable
[profile-<profile1name>]
name : <encoding_profile_name>
\description : <description> #optional
format : <format>
preset : <preset>
[streamprofile-<id>]
parent : <encodingprofile.name>[,<encodingprofile.name>..]
\type : <type> # "audio", "video", "text"
format : <format>
preset : <preset>
restriction : <restriction>
presence : <presence>
pass : <pass>
variableframerate : <variableframerate>- device/mp4target.gep (encoding_target_name=mp4target, encoding_profile_names=mp4, ...)
[GStreamer Encoding Target]
name=mp4target
category=device
description=MP4 (H.264, AAC) target
[profile-mp4]
name=mp4
type=container
description[c]=MP4 container profile
format=video/quicktime, variant=(string)iso
[streamprofile-mp4-0]
parent=mp4
type=video
format=video/x-h264
restriction=video/x-raw
presence=0
pass=0
variableframerate=false
[streamprofile-mp4-1]
parent=mp4
type=audio
format=audio/mpeg, mpegversion=(int)4
restriction=audio/x-raw
presence=0
[profile-(null)]
type=audio
format=audio/mpeg, mpegversion=(int)4
GstEncodingTarget
- gst_encoding_list_all_targets
GstPbutils.EncodingTarget
# create target with all profiles
# The name and category can only consist of lowercase ASCII letters for the first character, followed by either lowercase ASCII letters, digits or hyphens (‘-‘).
name = "mp4target"
category = GstPbutils.ENCODING_CATEGORY_DEVICE # "device"
description = "MP4 (H.264, AAC) target"
profiles = [container_profile, video_profile, audio_profile]
target = GstPbutils.EncodingTarget.new(name, category, description, profiles)# save target to ~/.local/share/gstreamer-1.0/encoding-profiles/<category>/<name>.gep
ret = target.save()
# list targets for all categories
category = None
target_list = GstPbutils.encoding_list_all_targets(category)
print("target_list: {}".format(target_list))
Encoding profile
(gst-validate-transcoding)
- general syntax of serialized encoding profile:
mux_format:[video_restriction->]video_format[+video_preset][|video_presence]:[audio_restriction->]audio_format[+audio_preset][|audio_presence]
- element factory:
- <muxer_factory_name>:<video_encoder_factory_name>:<audio_encoder_factory_name>
webmmux:vp8enc:vorbisenc- caps:
- <muxer_source_caps>:<video_encoder_source_caps>:<audio_encoder_source_caps>
- WebM (VP8 + Vorbis):
video/webm:video/x-vp8:audio/x-vorbis - MP4 (H.264 + MP3):
video/quicktime,variant=iso:video/x-h264:audio/mpeg,mpegversion=1,layer=3 - MP4 (H.264 + AAC):
video/quicktime,variant=iso:video/x-h264:audio/mpeg,mpegversion=4 - OGG (Theora + Vorbis):
application/ogg:video/x-theora:audio/x-vorbis - MPEG-TS (H.264 + AC3):
video/mpegts:video/x-h264:audio/x-ac3
- caps + preset:
- location of preset files
video/webm:video/x-vp8+youtube-preset:audio/x-vorbis"video/quicktime,variant=iso:video/x-h264+slow12mbps:audio/mpeg,mpegversion=1,layer=3""video/quicktime,variant=iso:video/x-raw,width=1920,height=1080->video/x-h264+slow12mbps:audio/mpeg,mpegversion=1,layer=3""video/quicktime,variant=iso:video/x-h264,width=1920,height=1080+slow12mbps:audio/mpeg,mpegversion=1,layer=3"- How to use the x264 encoding presets when rendering an XGES project
- caps + presence (number of times an encoding profile can be used inside an encodebin; 0: any):
video/webm:video/x-vp8|1:audio/x-vorbis- caps + restriction:
- ...:restriction_caps->encoded_format_caps:...
"video/webm:video/x-raw,width=1920,height=1080->video/x-vp8:audio/x-vorbis""matroskamux:x264enc,width=1920,height=1080:audio/x-vorbis"- loading profile from encoding target:
target_name[/profilename/category]/path/to/target.gep:profilename
GstEncodingProfile
GstPbutils.EncodingProfile
Container profile
GstEncodingContainerProfile
GstPbutils.EncodingContainerProfile
# container profile
name = "mp4"
description = "MP4 container profile"
#container_caps = "video/webm"
container_caps = "video/quicktime,variant=iso"
format = Gst.Caps(container_caps)
preset = None
container_profile = GstPbutils.EncodingContainerProfile.new(name, description, format, preset)
Video profile
GstEncodingVideoProfile
GstPbutils.EncodingVideoProfile
# video profile
#video_caps = "video/x-vp8"
video_caps = "video/x-h264"
format = Gst.Caps(video_caps)
preset = None
restriction = Gst.Caps("video/x-raw")
presence = 0 # allow any number of instances of this profile
video_profile = GstPbutils.EncodingVideoProfile.new(format, preset, restriction, presence)
container_profile.add_profile(video_profile)
Audio profile
GstEncodingAudioProfile
GstPbutils.EncodingAudioProfile
# audio profile
#audio_caps = "audio/x-vorbis"
audio_caps = "audio/mpeg,mpegversion=4" # AAC
format = Gst.Caps(audio_caps)
preset = None
restriction = Gst.Caps("audio/x-raw")
presence = 0 # allow any number of instances of this profile
audio_profile = GstPbutils.EncodingAudioProfile.new(format, preset, restriction, presence)
container_profile.add_profile(audio_profile)
Preset
- location of preset files
- /usr/share/gstreamer-1.0/presets/*.prs
- /usr/local/share/gstreamer-1.0/presets/*.prs
- /usr/share/pitivi/gstpresets/*.prs,
- ~/.local/share/gstreamer-1.0/presets/*.prs
- GST_PRESET_PATH
- ~/.local/share/gstreamer-1.0/presets/GstX264Enc.prs
[_presets_]
version=0.10
element-name=GstX264Enc
[slow12mbps]
speed-preset=slow
bitrate=12288
[slow700kbps]
speed-preset=slow
bitrate=700
- <profile_name>.prs
- ...
Gst.Preset
- location of target files (*.gep)
- gst-editing-services
- Exemples / Examples
- francesc.pinyol.m/python-gst-examples/ges
- Fun with video mixer (*)
- cut parts of a video using gstreamer/Python (gnonlin?)
- ...
- Resum / Summary
- Estructura / Structure (see Pitivi):
- pipeline
- timeline
- layer_1
- clip_asset_1.1
- clip_asset_1.2
- ...
- layer_2
- clip_asset_2.1
- ...
-
Python
headers
import sys
import gi
gi.require_version('Gst', '1.0')
from gi.repository import Gst, GLib
init Gst
class Player(object):
def __init__(self):
# init GStreamer
Gst.init(None)
init GES
# workaround to avoid "g_array_append_vals: assertion 'array' failed" when importing GES before Gst.init (using python3)
gi.require_version('GES', '1.0')
from gi.repository import GES
# init GES
GES.init()
main GLib loop
# create main glib loop
self.loop = GLib.MainLoop.new(None, False)
create timeline
# create timeline
timeline = GES.Timeline.new_audio_video()
create asset/clip
# create asset
asset = GES.UriClipAsset.request_sync(uri)
create layer in timeline
# create layer
layer = timeline.append_layer()
put clips in layer
# put clip in layer
# start=0.0
start_on_timeline = 0
# inpoint=60.0
start_position_asset = inpoint * Gst.SECOND
# duration=5.0
duration = duration * Gst.SECOND
clip = layer.add_asset(asset, start_on_timeline, start_position_asset,
duration, GES.TrackType.UNKNOWN)
create GES pipeline
# create GES pipeline
pipeline = GES.Pipeline()
connect message bus to callback
# connect bus messages to callback
bus = pipeline.get_bus()
bus.add_signal_watch()
bus.connect("message", self.on_message, None)
add timeline to pipeline
# add timeline to pipeline
pipeline.set_timeline(timeline)
(optional: only render)
ges_base_renderer.py
containers for output format
# container profile
name = "mp4"
description = "MP4 container profile"
#container_caps = "video/webm"
container_caps = "video/quicktime,variant=iso"
format = Gst.Caps(container_caps)
preset = None
container_profile = GstPbutils.EncodingContainerProfile.new(name, description, format, preset)
# video profile
#video_caps = "video/x-vp8"
video_caps = "video/x-h264"
format = Gst.Caps(video_caps)
preset = None
restriction = Gst.Caps("video/x-raw")
presence = 0 # allow any number of instances of this profile
video_profile = GstPbutils.EncodingVideoProfile.new(format, preset, restriction, presence)
container_profile.add_profile(video_profile)
# audio profile
#audio_caps = "audio/x-vorbis"
audio_caps = "audio/mpeg,mpegversion=4"
format = Gst.Caps(audio_caps)
preset = None
restriction = Gst.Caps("audio/x-raw")
presence = 0 # allow any number of instances of this profile
audio_profile = GstPbutils.EncodingAudioProfile.new(format, preset, restriction, presence)
container_profile.add_profile(audio_profile)
pipeline in render mode
# pipeline in render mode
pipeline.set_render_settings(output_uri, container_profile)
pipeline.set_mode(GES.PipelineFlags.RENDER)
progress
# progress
GLib.timeout_add(300, self.duration_querier, pipeline)
start
# start playing pipeline
pipeline.set_state(Gst.State.PLAYING)
self.loop.run()
stop
# unset
pipeline.set_state(Gst.State.NULL)
- gst-transcoding
- CLI tool: gst-transcoder-1.0
- Exemples / Examples
- Create a target file called device/mp4target.gep
- gst-transcoder-1.0 <input_file> <output_file>
- ...
- ...
http://www.francescpinyol.cat/.html
Primera versió: / First version: 27.X.2018
Darrera modificació: 5 de maig de 2019 / Last update: 5th May
2019
Cap a casa / Back home