Video quality with expert action

Will the video quality generated with expert action degrade because simulation data—consistently accompanied by expert actions during training—was introduced, while the current video quality evaluation is conducted under action-free conditions?