THFM: A Unified Video Foundation Model for 4D Human Perception and Beyond | Signal Canvas | ScienceToStartup