Hello,
I made video streaming system by using two jetson tx2 boards(one is server, the other is client) and gstreamer.
And I am using ximea camera.
I tested different resolution for video like 640x480, 1024x768 but there is no difference in latency.
In both cases, end to end latency is 50ms.
I wonder why?
Thank you.
Hi, I’m trying to do the same thing… Do you mind posting your commands to run on server and client respectively? I’m using the rtsp ./test-launch 'v4l2src device=/dev/video0 io-mode=2 ! image/jpeg,width=1280,height=720,framerate=60/1 ! videorate ! image/jpeg,framerate=12/1 ! nvjpegdec ! video/x-raw ! nvvidconv ! nvv4l2h264enc ! h264parse ! queue ! rtph264pay name=pay0 pt=96 audiotestsrc ! voaacenc ! queue ! rtpmp4apay pt=97 name=pay1' it has big latency…
Hi,
It would be interesting to measure how much of the latency that you observe is added by the capture alone, to determine which parts of the pipeline are contributing the most to the resulting latency.
The capture latency is usually tied to the capture framerate, so in @tedkim120221 case, I’d expect a similar capture latency in the TX2 if both resolutions used the same framerate.
Here are a couple of techniques to measure capture latency/results that might be interesting for you:
NVIDIA_Jetson_TX1_TX2_Video_Latency
NVIDIA_Jetson_TX2_-_VI_Latency_Measurement_Techniques
Jennifer
https://www.ridgerun.com/
Hi,
You may also consider to user jetson_multimedia_api. We have compared the two cases:
For your reference.