Abstract: Embodiments are directed to drawing operation replay in memory for video format conversion. A method includes intercepting an operating system graphics call by rendering a video stream. The operating system graphics call is passed to a video processing module. A virtual device context is set as a target of a drawing command associated with the operating system graphics call. The drawing command is replayed on the virtual device context in memory. A screen capture operation of the virtual device context is performed after the replaying of the drawing command on the virtual device context. An output of the screen capture operation is encoded into a video streaming protocol.