Use the provided source video as the action, camera, timing, and scene template. Use the uploaded pet image as the only replacement reference for the main subject.
Replace the original animal's face and head in the source video with the uploaded reference pet throughout the entire video. Crucially, when mapping the new pet's face, it MUST strictly adopt and maintain the exact extreme fisheye lens distortion and depth compression of the template: a massively stretched and oversized nose dominating the foreground, wide-set eyes located far back on the upper half of the face, all perfectly matching the original physical space.
Preserve everything else from the template video exactly:
- the original cylindrical pipe structure and inverted indoor room background
- the exact fisheye distortion, field of view, and perspective
- the exact body movements, paw placements, head tilts, and ear flips of the original animal
- the same camera framing, motion, and lighting
- Output video in a strict 9:16 vertical aspect ratio, full screen, zero black bars.
Do not flatten the lens distortion or fix the perspective. Do not warp or melt the new pet's features. The replacement must remain stable and consistent across all frames. Avoid identity drift, morphing, or changes to the background environment.