The other alternative to this is to do the conversion yourself, and this can often be faster than letting the driver do it for you. It’s certainly work benchmarking, as it’s the kind of code you can write up very quickly and simply.
I still could not make it work with glMapBufferRange() and without memcpy(), but I think I’ve implemented alternating buffers.
Here is my code: https://github.com/FilipVuk123/MyOpenGL/tree/main/forKhronos
Any further advice is welcome!
But this does not work when doing the same thing but with 3 cameras and 2 or even 3 alternating buffers per camera 
One always renders as a mix of all camera feed buffers.