How Kuaishou’s video model evolved from a text-to-video generator into a multimodal, controllable AI production system
From generative images to cinematic worlds