Use gstreamer to write on hdmi sink without dektop environment

Robotics & Edge Computing Jetson & Embedded Systems Jetson Xavier NX
1

Hi,

I want to use the nx for an embedded solution without the Desktop/GUI. The plan is to show a “live-view” on a HDMI Monitor. I need a “fast” booting solution and can’t wait for Gnome to start. Im using opencv with gstreamer and python or c.

Also I don’t want to have a gui for security reasons

Any Ideas ? Best regards Martin

3

Hi,
You may try nvdrmvideosink plugin. Please refer to examples in gstreamer user guide:
https://docs.nvidia.com/jetson/l4t/index.html#page/Tegra%20Linux%20Driver%20Package%20Development%20Guide/accelerated_gstreamer.html#wwpID0E0UP0HA

1 Like
4

thank you for this solution, works pretty nice:

gst-launch-1.0 filesrc location=1.mp4 ! qtdemux name=demux ! h264parse ! nvv4l2decoder ! nvdrmvideosink -e

with python3 /opencv i got: GStreamer warning: error opening writer pipeline: syntax error

ret, frame_read = cap.read()
frame_height, frame_width, _ = frame_read.shape 

gst_str = "appsrc ! videoconvert ! qtdemux name=demux ! h264parse ! nvv4l2decoder ! nvdrmvideosink -e "  
stream = cv2.VideoWriter(gst_str, 0, 25, (frame_width, frame_height), True)  

while True:
    ret, frame_read = cap.read() 
    stream.write(frame_read)

any ideas how to improve my pipeline ?

5

okay this pipeline works so far, im happy with it !

gst_str = "appsrc ! videoconvert ! nvvidconv ! video/x-raw(memory:NVMM),width=1920,height=1080,format=RGBA ! 
 nvdrmvideosink sync=false "

But if there any ideas how to improve this pipeline, i would be happy for some input

Best regards Martin

6

Hi,
There is BGR to RGBA conversion in videoconvert plugin, and copying data from CPU buffer to MVMM buffer in nvvidconv plugin. These utilize CPU usage so suggest execute sudo nvpmodel -m 2 and sudo tegrastats, to enable 6 CPU cores at 1.4GHz.