A model component that adapts its output based on the camera position and orientation (pose) to generate view-specific predictions.