Cinematic rescue POV video prompt with sound design

integrated_multimodal_description: 15 seconds, 16:9, realistic cinematic image-to-video. Use <Picture 1> as the only first frame and the only visual reference for the adult woman and the hangar. Preserve her face, hairstyle, goggles, earmuffs, dark olive open-collar work jacket, shoulder strap, sweat-sheened skin, dual mechanical arms, and the same dark steel-beam hangar with warm practical light on the left, long skylight at upper right, workbench, and distant background figures. The camera is first-person POV of a rescued person standing in front of her; only this person’s left forearm may appear briefly, and no second clear face is ever shown. Lock spatial continuity: she advances from center frame toward the camera. Upper-right frame and behind her right side hold the skylight and brighter corridor with cold daylight. Left frame remains the darker ceiling zone and safety passage. Danger falls from upper-right into the right corridor. She and the camera always retreat together toward screen-left, never swapping sides. Her left hand first pulls one earmuff aside to listen; her right hand grabs the rescued person’s left forearm and releases only at the end. Any camera turn stays on the same side of the action axis. Keep the palette and light stable: charred steel black-green, dark olive clothing, tungsten amber practicals, cool skylight white, and natural warm skin. Low saturation, deep olive and blue-gray shadows against amber highlights, roughly 3200K practicals against 5600K skylight daylight. High contrast with retained steel detail and soft highlight roll-off. Glass breakage creates only a brief cold-white flicker; time of day and overall color temperature never jump. [0.0-2.0s] Fast POV retreat. Continue directly from the frontal close view of <Picture 1>. From the first frame she takes a large step toward the camera, pulls one earmuff aside with her left hand to isolate a metallic tension sound overhead, and the POV retreats quickly by about one step. Two forceful footfalls drive visible vertical load through her upper body. The open collar and shoulder strap tremble with the same rhythm; sweat beads and loose hair catch warm side backlight. Close framing always keeps face and chest together. Her eyes first scan the person in front of her, then jump past the camera toward her own upper-right rear; her brow tightens slightly and her lips stop the breath she was about to release. [2.0-3.5s] Sudden slow-down, low-angle tilt up. She is still stepping forward slowly, never freezing. The camera tilts upward from the moving collar and shoulder strap to her eyes. In the curved reflection of the goggles, the bright skylight line in the upper-right trembles, and a speck of dust lands on her shoulder. Her eyes move first, then her head follows with a small turn. She blinks once quickly, bites down lightly in the back of the jaw, and returns attention to the rescued person. Do not reveal the full crack yet. [3.5-8.0s] Violent acceleration, lateral side-follow with one rebound deceleration. She leans toward the person and says in a low, short command close to one breath, <d>[English] Look at me.</d> The line is a command, not comfort. While speaking, her right hand reaches to the edge of frame and grabs the person’s left forearm. She lowers her center of gravity, drives off with her right foot, and pulls the person and the POV diagonally toward screen-left, out of the bright corridor, across about two steps. The camera is dragged by this force, sliding fast sideways and slightly downward, then settling under the darker left-side ceiling with one brief inertial rebound. The camera passes close to her upper body: the shoulder strap tightens, the open collar is tugged by wind and hard turn, and the chest makes a pronounced lateral swing caused by sideways acceleration and torso rotation. After the abrupt stop, it shows a smaller quickly decaying rebound consistent with real soft-tissue inertia and breathing. Loose hair whips opposite the movement and falls back across the shoulder, and the earmuff swings lightly. She checks first whether the person’s footing follows, then looks at the person’s eyes, keeping the mouth tight. [8.0-11.0s] Impact reveal, whip-pan. They have already stopped in the safe left-side zone. From beside her left shoulder, the camera whips toward the right-rear, but the endpoint still keeps part of her left shoulder silhouette in frame. At the skylight in the upper-right, one metal fastener snaps first, the frame bends, and a single pane loosens from one corner, tilts, strikes a beam, and shatters. Large shards fall first, then smaller fragments follow gravity into the bright corridor on the right, hitting, bouncing, and sliding outward. A little fine debris scatters toward the camera’s previous path and slows to a stop near the left-side ceiling edge. Cold white daylight passes through the fragments, and dust is thrown up by the impact, briefly veiling the frame before thinning. The distant background figures only turn, duck, or step back within their own positions. Do not turn this into a full hangar collapse. [11.0-15.0s] Stillness, fixed close shot. Use the impact dust as the transition back to the same safe left-side position. Hold a fixed half-body close shot. Her right hand still supports the person’s forearm and only lets go after confirming the person is steady. After the violent movement, her chest rises and falls visibly with urgent breathing. The open collar edge and shoulder strap tremble first, then settle gradually with one longer exhale. The composition contains both upper body and face, and attention rises with her movement: the jaw softens from tension, her lips part slightly, and she reestablishes eye contact with the person; one corner of the mouth loosens for only a brief instant. Then her eyes lift toward the upper-right skylight before the head follows, checking the danger again. End with both her and the camera still in the safe left-side zone, debris settled, and background figures continuing small evasive movement without freezing. Keep continuity exact throughout: same identity, same outfit openness, same goggles, earmuff, and shoulder-strap positions, same lighting direction, and the same factory depth. The camera must feel physically tugged by her grip. Hair, clothing, strap, skin, and chest motion must all follow step, torso rotation, hard stop, and breathing with proper inertia and decay. The glass falls only because of the failed upper-right skylight. All fragments follow gravity, collision, and occlusion. Face, fingers, ribcage, and cloth proportions remain stable. No extra foreground character enters. overall_soundscape: A continuous low industrial hangar ambience runs through the full 15 seconds: heavy room tone, faint metal resonance, distant footsteps, sparse background worker movement, and scattered machine hum. At the start, one sharp thin metallic tension ping is heard overhead; when she pulls one earmuff aside, the higher-frequency abnormal sound becomes clearer. During 0.0-2.0s, emphasize two heavy retreating footfalls, cloth and leather friction, close breathing, and light strap movement. From 2.0-3.5s, keep the tension thin and focused, with one tiny dust hit on fabric and the room tone narrowing as she listens. At 3.5-8.0s, she says once, <d>[English] Look at me.</d> in a low clear voice fully synchronized with the mouth. Add two urgent lateral steps, a light buckle tap from the shoulder strap, close breath, the grip on the forearm, and the camera-body pull. At 8.0-11.0s, the sound chain is precise: metal fastener break -> glass striking beam and shattering -> dense ground impacts and fragment sliding. After the impact, a brief ear-ringing tone compresses the sound field. In the last section, let that fade into her breathing as it changes from sharp to slower, with sparse distant reactions from the background figures. No extra dialogue, no voice-over, no subtitles. non_diegetic_music: Use a restrained low-frequency suspense pulse under the opening approach. Thin it out as she listens upward, then tighten it briefly during the grab-and-pull section. The glass-failure reveal should be driven mainly by physical sound, with only a short dark accent under the snap and fall. After the impact, drop the score back almost completely, leaving only a faint residual tension tone beneath her breathing and the far hangar ambience. Keep the music minimal and secondary to footsteps, breath, strap movement, metal failure, and falling glass.