Hello, I am not sure if this is not already being done, but recently I have managed to confirm that it is possible to offload reading of camera to the secondary M4 core and have VSYNC, which could allow for much higher speed in some examples, as the main core could do its own stuff without having to spend countless milliseconds reading the camera. So far I have only implemented this in Arduino using C++ but as its possible id imagine it should be possible with all STM32H747 boards.
We have full DMA offload in our firmware for reading image data. The main core gets 1 interrupt per frame and spends only a few microseconds setting up the frame capture.
The M4 core would only be useful for running some type of parallel lower power always-on audio detection or motion detection.