I did an implementation for capturing from an RTSP using FFMPEG (via
capture = CaptureFromIPCamera[
You can get this function from my
PacletInstall @ "https://github.com/arnoudbuzing/prototypes/releases/download/v0.2.9/Prototypes-0.2.9.paclet"
Or, if you prefer, you can just copy this one function from this source file:
The function works by simply calling the FFMPEG binary from the Wolfram Language with the right options to sample one image from an RTSP stream, and then importing that image. The implementation can probably be improved a lot, for example by directly importing from FFMPEG's standard output.