So, I am building a project based on your Argus cudaBayerDemosaic example which uses RAW_16 format. From the example, I found out that the values are 14-bit based, even when my sensor is a 10bit sensor.
My project is working as expected, but before going into production I would like to know more about the RAW_16 format since there is not much documentation. Why there is this bit shift, what happens with the data (is there any more processing involved? Do I have any kind of altering of the original data or even losing depth in some processing step?).
Will I get the same data as with V4L2, but just bitshifted?
As far from my tests, even auto exposure seems to work fine like this.