Tool/software:
The thread below has useful information on where each part of the vision pipeline is normally handled (R5 vs A53).
My question is: if we wish to have the fastest possible presentation of video from one MIPI camera to the display, prior to Linux booting, could this be fully implemented on R5 including frame handling? Is there a technical or practical limitation to prevent that?
Or is this better to handle in the QNX way, where the necessary drivers are loaded on R5 and A53 respectively, but prioritized ahead of the rest of the system.
Our target for video to the screen is 1 second from power on.