La Calabaza Dance AI Video Generator

Recast the selected 14-second scene. The reference video controls exact movement, framing and edit timing; the intervals below describe the intended beats.

CAST — CENTRE · COWBOY HAT: @Image1 supplies ONLY the centre dancer wearing the cowboy hat. Keep this identity from opening to final frame.
LEFT DANCER: Keep the original cream kitten on the viewer’s left unchanged in @Video1. This role NEVER inherits an uploaded identity.
RIGHT · GLASSES: Keep the original glasses-wearing kitten on the viewer’s right unchanged in @Video1. This role NEVER inherits an uploaded identity.

IDENTITY PRIORITY: Uploaded role photos are the ONLY authority for the assigned characters, including the ending. Environment and pose references must never supply a face. If a role photo is a cat or dog, make that character an upright anthropomorphic animal FROM THE FIRST FRAME: its own furry body, two furry forelimbs ending in paws, two hind legs, its own tail, head, ears and muzzle. Never begin as a four-legged animal then morph into a human body. Never use bare human hands, skin-coloured fingers or human forearms on an animal. When both role photos are animals, BOTH protagonists remain those two animals in EVERY scene, including flashbacks; no human protagonist may appear.

REFERENCE MAP:
Identity photos only; the original video supplies the environment and motion.

SETTING: The three-kitten night roadside in @Video1, lit by a nearby phone light and distant street lamps. Keep the ground, shadows, three subject positions and low vertical camera.
SELECTED MODE: Cowboy lead

CONTINUITY: Use @Video1 for the real La Calabaza routine, not a generic country line dance or a pumpkin costume. Exactly THREE dancers remain visible: replace only the assigned roles and keep all unassigned source kittens. Bind identities to their bodies through turns; never swap uploaded pets when they overlap. Keep the central cowboy hat and right-side glasses attached to their original roles. Each uploaded pet keeps its own coat pattern, muzzle, ears and tail, upright with two hind legs and exactly two forepaws. Do not add human hands, hooves or boots. Do not clone one upload into unassigned roles.

SHOT 1 — 0–3 seconds: establish the original three-person formation and follow the synchronized short side steps and alternating forepaw swings.

SHOT 2 — 3–6 seconds: match the source hip bounces and forepaw accents, keeping all three identities and head accessories stable.

SHOT 3 — 6–10 seconds: follow the source turn and back-facing beat, maintaining the same left-centre-right geography and natural tails.

SHOT 4 — 10–14 seconds: return toward the camera and complete the source forward paw reach and final dance accents. Keep three separate bodies and avoid merged paws.

FINISH: Keep exactly the assigned subjects throughout. No identity drift, cloned bodies, extra limbs, costume changes, captions, logos, watermarks or unrelated actions. Deliver a silent MP4; music is added by the user in their social app. Small motion differences are acceptable; preserve the action order, shot contents and ending.