Maintaining visual consistency across consecutive shots has historically been the central bottleneck of generative video. Early text-to-video platforms produced impressive standalone clips, but attempts to stitch those clips into a cohesive narrative resulted in morphing faces, shifting outfits, and fluctuating environment details. For production teams, these continuity errors rendered generative tools impractical for narrative storytelling…