I’m trying to make a program that will capture the screen on both Linux and windows and then encode the output video on the fly. Because the capture SDK is only supported on Linux (and not all Nvidia GPUs are compatable) I wanted to separate the capturing and encoding. Not only that but there will be preprocessing before encoding.
So how should I go about capturing the screen and then using the NvEncode library in a decoupled way? I can’t use system memory, only GPU memory. I’m sure I’m missing something.
I’m providing no code examples of my attempts because I’m embarrassed and will certainly start again from scratch.
Thanks in advance