On DM368 we are using the IPIPE/RSZ hardware scalers to generate two size of output images. We would like to create more resolutions without stopping the ISP and sending data back to the IPIPE/RSZ.
Doing this as a software scale in C is very painful for the CPU, but we have been told that it is possible to accelerate using the IMX.
Please could you explain how we go about doing horizontal/vertical averaging using the IMX?
Many thanks.