AI 视频生成器

IMAX Reaction

模型

提示词

Vertical smartphone video, shot on iPhone 15 Pro, 4K HDR, 60fps, realistic handheld shake, natural low-light sensor noise, authentic UGC found-footage style. Filmed inside a massive IMAX 70mm auditorium at night: enormous wall-to-wall IMAX screen in tall 1.43:1 aspect ratio, steep stadium seating with dark red-and-black recliner seats, amber aisle floor lights, dim red EXIT signs. The screen is an enormous curved front screen at the bottom of the hall, dominating the auditorium like a true IMAX 70mm venue. The only light in the room is the glow of the screen itself.

**GEOMETRY LOCK — the screen is ALWAYS in front:** The auditorium has ONE screen, located at the FRONT of the hall, at the bottom. The filmer sits in the upper rows facing the screen — the screen is always IN FRONT of the filmer, downhill from the seat. BEHIND the filmer there are only the upper rows of seats, the back wall and the exit doors. There is NO screen behind the filmer, no second screen, no glowing wall behind — in the selfie shot, the background behind the filmer must show only dark rows of seats with other viewers and the back of the auditorium. The glow on the filmer’s face comes from in front, from the screen being faced.

**The filmer:** use the uploaded face photo <<<image_1>>> as the identity reference for the filmer, WITHOUT glasses — no glasses on the face at any point — wearing a plain black T-shirt. If the uploaded face photo is male, make the filmer a young man and make the whispered voice masculine; if the uploaded face photo is female, make the filmer a young woman and make the whispered voice feminine. The ENTIRE video is shot from the filmer’s single phone — the same phone that is recording this footage. That phone is NEVER visible in frame because it IS the camera we see through. CRITICAL: the filmer holds NOTHING in the hands. No second phone, no other smartphone, no device, no object of any kind ever appears in the hands or in front. The hands are completely empty (they only hold the recording phone, which stays out of frame). In the selfie shot the hands are NOT visible at all — only the face and shoulders. NEVER show a phone screen, a phone body, or any rectangular device anywhere near the filmer. The filmer never stands up and never leaves — stays seated in the seat the entire time through all the chaos, only turning and panning the phone.

**CRITICAL — two separate worlds divided by the screen:** Everything in the movie exists ONLY as a flat 2D projected image on the screen surface. The battle, the fortress, the fire — none of it is physically in the room; it is light projected onto a huge flat fabric screen, seen at an angle from the filmer's seat, its glow spilling onto the first rows. The auditorium is real three-dimensional space. The projected image looks like a movie — film grain, halation, cinematic color — while the auditorium looks like raw iPhone footage. Two separate layers, divided by the physical fabric of the screen, until the exact moment it tears.

**Beat 1 — Selfie FIRST (strict order):** The video MUST START with the front-camera selfie shot. Low-angle selfie from the seat, upper center rows. The filmer films THEMSELF — only the face and shoulders in frame, NO glasses on the face, the hands NOT visible, lit by the flickering orange glow coming from the screen IN FRONT. Behind in the background: only dark upper rows with other viewers eating popcorn, and the back wall of the auditorium — NO screen visible behind. The filmer glances toward the screen (off-camera, in front) for half a second, leans slightly closer to the lens and WHISPERS with a grin, in a hushed excited movie-theater voice, barely above breath so the people around don't hear: *"This... is... absolute cinema! It's an IMAX!"* — a quiet conspiratorial whisper, close-mic sound of whispered speech picked up by the phone mic over the booming movie audio, popcorn crunching nearby. ONLY AFTER the whispered line does the filmer rotate the camera 180° forward, away from themself, toward the screen at the front of the hall. This is the only moment the filmer is visible.

**Beat 2 — On-screen sequence (natural, chaotic army):** After the camera turns to the screen: autofocus hunt, realistic digital zoom. The huge screen is a glowing flat rectangle at the front of the dark hall, first rows silhouetted against it.
ON the screen, the movie plays in this order:

1. Static night shot: a massive ancient sandstone fortress wall with sharp pointed crenellations, two burning torches on the wall, low fog, closed heavy dark wooden gates.

2. A few soldiers push the heavy gates open — the doors swing inward unevenly, one door faster than the other, revealing a burning city glowing behind.

3. An army pours out of the gateway and charges toward the viewers — **naturally and chaotically, like real battle footage: an irregular loose crowd, NOT a neat formation, NOT symmetrical rows, NOT synchronized.** Soldiers run at different speeds, overtaking each other, staggered and scattered, some stumbling on the uneven ground, shields and spears held at different angles, gaps and clusters in the crowd, some men shouting, dust kicked up unevenly. Front runners fill the frame at different moments, not in one line. Documentary-style war chaos, like a real historical battle scene — natural human running, varied body types and armor details, no mirrored or cloned soldiers, no warping figures.

All of this stays flat cinematic footage contained inside the screen rectangle — film grain, halation.

**Beat 3 — The fabric tears (KEY MOMENT):** The charging soldiers on screen reach full size — and the tear begins.

*Realistic fabric physics:* The screen is heavy stretched projection canvas under tension. One spot BULGES outward — a sharp push from behind, the projected image warping over the bulge. A spearhead punctures it — a small split. The tear RIPS along the fabric in fast jagged lines with a loud deep ripping sound of heavy canvas, propagating from the puncture outward and downward. Torn flaps and strips sag, swing and flutter, loose threads at ragged edges, fabric recoiling as tension releases. NOT like breaking glass or crumbling wall — exactly like heavy woven cloth: bending, folding, flapping, hanging.

*Light and smoke:* the instant the hole opens, warm amber-orange light blasts out from behind the screen — fire and torch glow — a volumetric flickering shaft into the dark hall, thick smoke pouring from the tear, rolling low over the first rows, embers drifting. Briefly two light sources: the dying projection on the torn strips, and the live fire-glow from the hole. Then the projection cuts off — house lights snap on, harsh white light mixing with orange glow and smoke. Phone reacts: blown-out exposure at the flash, abrupt white-balance shift, autofocus hunting through smoke. About 10 real physical armored warriors climb through the tear — silhouettes backlit first, then visible in house light. Some grab the torn edges and rip the flaps wider. They spread chaotically: some run between rows, some charge up the aisle stairs, some stop and scan the hall. No breaking walls, no debris — ONLY torn canvas, light and smoke.

**Beat 4 — Audience reacts (filmer stays put):** The filmer NEVER leaves — stays seated in the upper-row seat and keeps filming, panning around from a fixed position: the hall, the torn screen hanging in strips, soldiers on the stairs. The audience reacts unevenly: some viewers from the middle rows bolt backward toward the exits behind, running up the stairs past the seat; others stay frozen sitting in the front rows, too stunned to move, heads turning; someone crouches between seats. Genuine startled reactions on the phone mic: "oh shit...", "what the hell is that?!", scattered screams, seats flipping up, popcorn spilling. The camera shakes but recording continues from the seat.

**Beat 5 — Agamemnon enters and IMMEDIATELY does the dance:** Thin smoke drifts from the tear. Out of the smoke, Agamemnon steps through — wearing near-black bronze ancient Greek armor with a polished but battle-worn surface, a tall black horsehair crested war helmet, the face fully hidden in shadow behind the helmet, broad dark chest armor with two round gold medallion clasps at the shoulders, black underlayers, armored skirt panels, a long black cape, and a sword sheathed at the side. He walks a few steps in, stops in front of the torn screen at the bottom of the hall, facing the auditorium — and IMMEDIATELY, without pause, locks his gaze on the camera high up in the upper rows and starts dancing, deadpan, staring into the lens the entire time.
The dance <<<video_1>>> recreate this reference: motion control! The contrast between the goofy bouncy dance and his instant return to statue-like menace is the punchline. Throughout the entire dance: helmet stays on, face stays hidden in shadow, sword stays sheathed at his side, the black cape swings naturally with every rock and bounce. He never looks away from the camera. Filmed from a long distance from the filmer’s seat, shaky handheld digital zoom pushing in, realistic zoom quality drop and micro-shake. On the phone mic: a stifled laugh-gasp, someone nearby whispering "...is he dancing?!", distant murmurs.

**Style tags:** photorealistic, raw iPhone HDR look, no color grading on phone footage — 70mm film look ONLY inside the projected image, 60fps motion blur, realistic autofocus breathing, live clipped phone-mic audio, ultra-realistic people, POV strictly from the single recording phone — the recording phone IS the camera and is never visible, NO second phone, NO phone in the filmer’s hands, NO device in frame, the filmer’s hands always empty, hands not visible in the selfie shot, the filmer wears NO glasses, speaks only in a hushed whisper (movie-theater etiquette), whispered close-mic voice over loud movie audio, the filmer never stands or leaves the seat, the screen exists only at the FRONT of the hall — never behind the filmer, chaotic irregular army crowd on screen — no neat formations, Agamemnon starts dancing immediately upon entering and stays deadpan, the video opens with the selfie before turning to the screen.

10197 / 50000

参考图片（可选）

画面比例

时长

分辨率

生成音频2 倍积分Basic+

所需积分50

视频预览

加载中...