WorldCompass teaches video world models to follow actions better and keep pictures pretty by using reinforcement learning after pretraining.
This paper creates MotionEdit, a high-quality dataset that teaches AI to change how people and objects move in a picture without breaking their looks or the scene.