Duration: 30 seconds, Single continuous shot, photorealistic, cinematic.

Duration: 30 seconds, Single continuous shot, photorealistic, cinematic. SOURCES — STRICT SEPARATION: - IMAGE 1 = the ONLY source of visual/photographic content (100%). Every pixel of appearance in the output derives from Image 1: the subject and the entire scene, including background. - VIDEO 1 = motion source ONLY. It contributes no appearance, no identity, no background, and no color/lighting — only movement, timing, and camera behavior. CRITICAL — CLEAN RENDER, ZERO ANNOTATIONS: The final video must contain zero graphical annotations of any kind — no grid lines, guide lines, tracking lines, crosshairs, registration marks, wireframes, mocap or skeletal overlays, watermarks, timecodes, or text overlays. Any such marks appear in the references only as production guides; render only the clean underlying photographic content. IDENTITY (from Image 1, locked): The subject is defined entirely by Image 1 — face, bone structure, skin, eyes, hair, body type, age, and wardrobe exactly as shown. Keep this appearance locked and stable from the first frame to the last: no morphing, no feature drift, no blending with any person from Video 1, even during fast or extreme motion. The performer on screen is always the subject from Image 1. PERFORMANCE / MOTION (from Video 1, transferred exactly): Take the motion of the subject in Video 1 and apply it exactly to the subject from Image 1. This includes all body movement, gestures, posture, weight shifts, facial expressions, micro-expressions, blinks, eye movement, lip movement (fully lip-synced), timing, and pacing. Transfer the performance, not the performer — replicate Video 1's movement precisely while the appearance stays 100% Image 1. BACKGROUND — ANIMATE THE IMAGE 1 SCENE: Bring the background of Image 1 to life with natural, physically plausible motion so the frame never looks like a still with a moving cutout. Every applicable background element should move believably and continuously — e.g. swaying foliage, drifting clouds, rippling water, moving light and shadow, blowing fabric/hair, ambient crowd or traffic motion, dust, steam, or reflections — appropriate to what actually appears in Image 1. This motion is generated for realism and ambience; it is not copied from Video 1's background. Keep it subtle, coherent, and grounded — pleasing and realistic, never chaotic or distracting. CAMERA (from Video 1, matched shot-for-shot): Follow Video 1's camera exactly — replicate every pan, tilt, dolly, truck, zoom, push-in, pull-out, handheld sway, roll, and shake with identical direction, speed, and timing, so the virtual camera tracks the reference frame by frame. Match Video 1's framing, lens, and depth of field. FOREGROUND OBJECTS & OCCLUSION: Any objects in front of the subject are subject to the same camera motion — they must move, parallax, and reframe with the camera in a physically grounded way, maintaining correct relative position, scale, and perspective throughout. The subject must never pass through, clip, overlap, or glitch through foreground objects; occlusions stay physically accurate, with objects realistically blocking or revealing parts of the subject as the camera and performance dictate. Keep every object stable and properly anchored. AUDIO: Follow Video 1's audio track precisely, fully synced to the visuals. OUTPUT: Cinematic, photorealistic, sharp focus, natural physics, consistent lighting anchored to Image 1, clean uncluttered frame.