I’m trying to port over our product pipeline from the NXP IMX8 (Toradex Carrier) to the Jetson Line.
We are trying to achieve a seamlessly seeking video player, and have done so very successfully using the IMX8 (Toradex BSP).
We are Utilising gstreamers segment seeking usecase to jump between different segment of a H264 encoded .mp4
I’m currently trying to port the pipeline we currently use, and have attempted the following:
filesrc ! qtdemux ! h264parse ! queue ! nvv4l2decode ! queue ! glimagesink
However it appears that “nvv4l2decode” just locks up when performing a seek with “SEGMENT_SEEK” flags.
“SEGMENT_DONE” appears on the bus almost instantly after the seek event is pushed
The issue does not occur when using the same pipeline but instead of nvv4l2decode, avdec_h264, however we want the hardware decoding, as avdec_h264 really pushed the CPU.
I have also tried using nv3dsink, but with no luck.
I’ve tried this on an Orin Nano and Orin NX but with the same issue. We are up to date with our jet-packs.
Is there something I’m missing?
I’d like to to use the Jetson line as we’ve had success previously with the TX2 but this is a critical hurdle in the production pipeline.
Thanks in advance.
William