recording event with Xavier

[Andrey1984]

With TX2 I would use :

[ Use case II-b-1: noaudio]  
$ gst-launch-1.0 nvcamerasrc fpsRange="30.0 30.0" ! 'video/x-raw(memory:NVMM), width=(int)1920, height=(int)1080, format=(string)I420, framerate=(fraction)30/1' ! omxh264enc ! qtmux ! filesink location=test1.mp4 -e
[ Use case II-b-2: noaudio]   
$ gst-launch-1.0 nvcamerasrc ! 'video/x-raw(memory:NVMM), width=(int)1920, height=(int)1080, format=(string)I420, framerate=(fraction)30/1' ! omxh264enc ! 'video/x-h264,stream-format=(string)byte-stream' ! filesink location="test.h264" -e

and that would record video with onboard CSI camera.
How the sequence will look like for Xavier?

[Honey_Patouceul]

With Xavier and R31, you would use nvarguscamerasrc instead of nvcamerasrc and NV12 format instead of I420, such as:

gst-launch-1.0 <b>nvarguscamerasrc</b> ! 'video/x-raw(memory:NVMM), width=(int)1920, height=(int)1080, format=(string)<b>NV12</b>, framerate=(fraction)30/1' ! omxh264enc ! 'video/x-h264,stream-format=(string)byte-stream' ! filesink location="test.h264" -e

[Andrey1984]

yes, thank you.
And how to add voice to it?
something like below?

! queue ! mux. alsasrc num-buffers=1000 device="hw:2,0" ! voaacenc ! queue ! qtmux name=mux !

[Honey_Patouceul]

The following should work:

gst-launch-1.0 -e nvarguscamerasrc ! 'video/x-raw(memory:NVMM),width=1920, height=1080,format=NV12,framerate=30/1' ! omxh264enc ! 'video/x-h264,stream-format=byte-stream' ! h264parse ! queue ! muxer.video_0           alsasrc device="hw:2,0" ! voaacenc ! queue ! muxer.audio_0    qtmux name=muxer ! filesink location=test.mov

Note if you want to record a given amount of time that num-buffers for nvarguscamerasrc is frames number, while for alsasrc the unit is 10ms, so for recording 10s you would try:

gst-launch-1.0 -e nvarguscamerasrc num-buffers=300 ! 'video/x-raw(memory:NVMM),width=1920,height=1080,format=NV12, framerate=30/1' ! omxh264enc ! 'video/x-h264,stream-format=byte-stream' ! h264parse ! queue ! muxer.video_0           alsasrc num-buffers=1000 device="hw:2,0" ! voaacenc ! queue ! muxer.audio_0    qtmux name=muxer ! filesink location=test.mov

[Andrey1984]

Thank you Honey_Patouceul,
With your help and Xavier I will take the niche of recording of events consequently. 2 hours of recording take approximately 3GB [ 1920x1080 resulution] ov5693

[Andrey1984]

upon checking it seems that maximum is width=2592,height=1944 resolution mode

gst-launch-1.0 nvarguscamerasrc ! 'video/x-raw(memory:NVMM), width=(int)2592, height=(int)1944, format=(string)NV12, framerate=(fraction)30/1' ! omxh264enc ! 'video/x-h264,stream-format=(string)byte-stream' ! filesink location="test.h264" -e

[Andrey1984]

@Honey_Patouceul,
May be you could help with another concern?
A remote colleague who has PI asked to approach figuring out if the sequence that they are using could bring better quality with it.
It turned out that they are using :

gst-launch-1.0 -v v4l2src ! video/x-h264, framerate = 15/1, width=1280, height=800 ! h264parse ! rtph264pay config-interval=1 pt=96 ! udpsink host=10.0.0.2 port=900

I never used PI, before, but I managed to assemble test-launch at it with a purpose of approaching rtsp streaming with it as I done with jetsons before, but it seems a challenge to start it in a workeable way.
:

sudo apt-get install libgstrtspserver-1.0 libgstreamer1.0-dev
wget https://gstreamer.freedesktop.org/src/gst-rtsp/gst-rtsp-server-1.14.1.tar.xz
tar -xvf gst-rtsp-server-1.14.1.tar.xz
cd  gst-rtsp-server-1.14.1
cd examples
gcc test-launch.c -o test-launch $(pkg-config --cflags --libs gstreamer-1.0 gstreamer-rtsp-server-1.0)

loading kernel module

sudo rpi-update # to make sure the v4L2 drive is available.
sudo modprobe bcm2835-v4l2 # to load it and create /dev/video0

at two terminals of that one is sending and one approaches receiving
it seems that works at rpi is

raspivid -o - -t 0 -w 1280 -h 720  -fps 25 -b 1000000 | cvlc -vvv stream:///dev/stdin --sout '#rtp{sdp=rtsp://:8554/}' :demux=h264

and that can be played from the very same device with

cvlc rtsp://127.0.0.1:8554/

Testing the latter stream via recording it at receiver side with:

/usr/bin/vlc -vvv rtsp://192.168.1.128:8554/ --sout=file/ts:/media/path/to/save/location/recording-$(date +"%Y%m%d%H%M%S").ts -I dummy --stop-time=480 vlc://quit

Upd:
improved resolution

/usr/bin/vlc -vvv rtsp://192.168.0.124:8554/ --sout=file/ts:/home/nvidia/recording-$(date +"%Y%m%d%H%M%S").ts -I dummy --stop-time=4800 vlc://quit

passing test screen works:
streaming

./test-launch "videotestsrc ! videoscale ! "video/x-raw,width=720,height=576,format=I420" ! rtpvrawpay name=pay0 pt=96 "
stream ready at rtsp://127.0.0.1:8554/test

receiving

gst-launch-1.0 rtspsrc latency=50 location=rtsp://127.0.0.1:8554/test !  rtpvrawdepay ! videoconvert ! ximagesink -v

[Andrey1984]

Finally:

raspivid -t 0 -w 1280 -h 720 -fps 25 -g 5 -b 4000000 -vf -n -o - | gst-launch-1.0 -v fdsrc ! h264parse ! gdppay ! tcpserversink host=127.0.0.1 port=5000

then

./test-launch "( tcpclientsrc host=127.0.0.1 port=5000 ! gdpdepay ! rtph264pay name=pay0 pt=96 )"

seems to provide a readable output at https://127.0.0.1:8554/test
Will try to get the rtsp stream to integrate into deepstream xavier

[Andrey1984]

Guys,
May be you know to capture an single image from CSI camera at Xavier? I should have asked the question in the past, but seems that either I have forgotten how to do that or never got a response at forum.
If I want to take an image captured from a stream, what shall I do?
At TX2 there were keys that worked for recording of something while gstreamer dialog window is open as far as I remember.
As for now I have to use PrintScreen, but that obviously reduces resolution and adds border like in the attached image.
Then I am trying to apply Neural style, and I am just wondering how it would look like without borders.
original
External Media
stylized
External Media
stylized 2
External Media

[Honey_Patouceul]

You may try:

gst-launch-1.0 nvarguscamerasrc num-buffers=1 ! nvvidconv ! jpegenc ! filesink location=camerashot.jpg

[Andrey1984]

Thanks!
It takes persistent practice to have in memory all these variables.

[Andrey1984]

with higher resolution it will probably be:

gst-launch-1.0 nvarguscamerasrc num-buffers=1 ! 'video/x-raw(memory:NVMM), width=(int)2592, height=(int)1944' ! nvvidconv ! jpegenc ! filesink location=camerashot.jpg

Is it the best the camera can do e.g. for a portrait?