What problem does it solve? AI video models cap each generation at a few seconds, so creators cannot produce longer clips or multi-shot narratives from a single seed video. This Skill continues an existing Veo clip past its per-call duration limit while preserving motion, lighting, and subject identity. ## Core Features & Use Cases - Endpoint routing: Chooses between Veo 3-1 Extend (hero quality) and Veo 3-1 Fast Extend (lower cost for iteration) based on quality versus budget intent. - Prompting guidance: Ships documented patterns for camera anchoring, one-beat-per-extend structure, and chaining outputs to build shot-by-shot narratives. - CLI execution: Invokes runcomfy run google-deepmind/veo-3-1/extend-video with a source video URL and continuation prompt, then downloads the result. - Use Case: Take an 8-second text-to-video clip of a character, extend it to 16 seconds with a camera push-in, then chain two more extends to build a three-beat story. ## Quick Start Ask the agent to extend your video by providing the clip URL and a short prompt describing what should happen next, for example: extend this video so the camera slowly pushes in while the character turns toward the window.
