Server setting to fall behind instead of dropping frames?

Answered

Greg Akselrod

March 11, 2026 00:47 Edited

We have a machine vision system that connects to the VMS server to retrieve an RTSP stream from cameras. This server sometimes hits max CPU and (we think) the RTSP stream (pass through) is dropping frames, because the VMS's saved footage doesn't have any artifacts/corruption, but the stream we are processing does. Is there a way to configure the server to prioritize passing through all the frames even if it can't keep up with real clock time (and ideally catch up by sending all the frames quickly)? Running 6.0.5.

0

• 

  Ichiro

  • Network Optix team

  March 11, 2026 23:52

  Hi Greg,

  Thank you for your question. Unfortunately, this configuration is not supported at the moment.

  Based on your description, the issue may appear to be related to CPU spikes. As a first step, it would be helpful to investigate the root cause of the CPU usage increase, ex: you are pulling a transcoded RTSP (change resolution of stream, i.e. not the native of primary and secondary stream)

  In general, we do not always recommend using RTSP for real-time analytics, especially if your use case requires low latency and fast frame delivery. In such scenarios, we would suggest using the metaSDK (C++) to retrieve frames directly from the Media Server, either in compressed or uncompressed format, depending on your requirements.

  Thanks.

   

  0

Please sign in to leave a comment.