MotionLCM, a real-time controllable motion latent consistency model, can generate high-quality human motions with text and control signals efficiently.
