Currently the Camera API grabs software frames, but it's possible to get 0-copy textures instead. Which is faster if you only want to render the frames.
So that return value is a 0-copy texture for the SDL_Renderer.
( some example to use 0-copy textures with cameras: https://github.com/libsdl-org/SDL/pull/8565
testffmpeg.c also shows 0-copy textures creations and update. )
I would probably return properties that can be used to create a texture or interop with some other display mechanism, rather than tie it directly to an SDL renderer.
Currently the Camera API grabs software frames, but it's possible to get 0-copy textures instead. Which is faster if you only want to render the frames.
suggestion is to add:
So that return value is a 0-copy texture for the SDL_Renderer. ( some example to use 0-copy textures with cameras: https://github.com/libsdl-org/SDL/pull/8565 testffmpeg.c also shows 0-copy textures creations and update. )