TASK AND REFERENCE PRIORITY Generate an exactly 24.00-second, landscape 16:9, full-frame photoreal

Prompt

TASK AND REFERENCE PRIORITY Generate an exactly 24.00-second, landscape 16:9, full-frame photoreal fantasy film with a live-action-style VFX finish. It should look like practical location photography seamlessly combined with cinematic visual effects, not a game cutscene. Use the original creature and branded city references to create believable weight, golden daylight, immense scale and one continuous chain of destruction. Follow the shot timings below. Do not globally speed up the action or recreate a named film or anime character. @img1 = our original creature character board. Four panels, left to right: full-body front, full-body left profile, full-body back, frontal head-and-shoulders close-up. All four depict ONE creature, not four creatures. Lock its intact iron-grey scaled hide, ochre keratin brow and shoulder/back plates, broad short nonhuman face, thick neck, barrel torso, long arms and three-toed feet. Change poses naturally for the action. Do not reproduce board dividers, studio background or panel layout. @img2 = our matching PhotoGPT engraved-wall city reference. Use it for the city, architectural scale, golden daylight, the composition between the domed bell tower and pointed tower, and the brand location on the inner central wall. The creature must be the same individual as @img1; @img1 takes priority for close-up anatomy. @img3 = the correct PhotoGPT horizontal wordmark on its padded canvas. Use it only for letterforms, aperture blades, spelling, spacing and proportions. Do not copy its black background into the scene. References constrain design, silhouette, colour, city composition and logo shape, NOT a studio-rendered or game-rendered finish. The photographic direction below takes priority for faces, materials, hair, lighting and camera behaviour. Transform the creature design into a believable living being and the city reference into a real exterior with atmospheric depth. Do not copy studio illumination or uniform render sharpness. Use only these three references for this version. Do not mix in earlier character boards, film screenshots or source videos. References are visual constraints, not mandatory first frames: open with the aerial city view. Animate a continuous three-dimensional world, not still-image zooms, a slideshow or an animated reference board. There is no human character board. Generate civilians from the text descriptions below; do not give them the creature's face or skin. PHOTOGRAPHIC STYLE, MATERIALS AND CAMERA Everything has the appearance of live-action cinema: a crew filming real medieval streets, with a living giant and destruction integrated through invisible photoreal VFX. Civilians look like real adult extras; buildings look like physical stone, timber and plaster. The creature has skin elasticity, joint compression and immense body weight. Match exposure, lighting direction, atmospheric perspective, focus, motion blur and subtle grain across every element. No sharply pasted-on creature or debris layers. Human faces have natural asymmetry, pores, slight skin-colour variation, real eyelids and restrained moist eye highlights. Hair has individual strands and wind-driven flyaways. Clothing has fibres, seams, folds, dust and believable weight. No waxy smoothing, plastic hair clumps, generic digital faces or cloned crowds. All people are newly generated fictional adults, not imitations of particular real people. Creature skin has varied roughness, pressure folds, small scales and accumulated dust. Keratin growth ridges and worn edges vary in scale; the whole animal must not look like stone or metal. Its body casts directionally correct shadows on the wall. Fingers press against masonry and feet compress against the ground. Steam and dust respond to light and occlusion; distant detail fades naturally through the atmosphere. Maintain warm golden daytime sun, yellow-orange backlight, grey-green shadows, iron-grey skin, muted ochre keratin and thick sunlit clouds. Keep sunlight direction consistent in world space. Vapour edges and fine dust catch the backlight. No night, snow, neon, magical beams, lightning transformations or unexplained fire. Later grey-brown dust progressively obscures the sky without changing time of day. Use the optical character of real cinema cameras and spherical lenses. Target natural 24fps, approximately 180-degree-shutter motion, subtle grain and soft highlight roll-off. Avoid oversharpening, exaggerated clarity, HDR halos and artificial lens flares. Sun and sky motivate the illumination; preserve shadow detail without flattening it with studio fill. White balance and exposure remain coherent between cuts; breach sunlight retains detail rather than clipping into a large white patch. Approximate full-frame-equivalent lenses: 24–35mm for aerials and wide city views; 35–50mm for the low wall-top face/hand shot; 50–70mm for the tower-framed architectural view; 28–40mm for street action and debris tracking. Hold focal length steady within each shot. No digital zooms, fisheye exaggeration, arbitrary distortion or breathing zoom. Do not change the required composition merely to create background blur. Wide shots retain enough depth of field to read city, wall and creature scale. Close shots have distance-appropriate defocus. Foreground grass and passing debris may briefly fall out of focus, but main actions and the readable-logo window remain focused. No blanket ultra-shallow depth of field or miniature-city appearance. Camera movement has physical mass and inertia. The opening feels like a high aerial or crane easing downward; the branding view is a stable, very gentle push; street shots feel like controlled operator tracking. Use short shakes only at impacts or urgent avoidance, then recover. No constant random shake, weightless game-camera orbit, movement through buildings or instantaneous repositioning. Movements start and stop naturally. Create scale through the giant face, near hand, low grass, lifted sole and foreground towers. Slow shots still contain weight shifts, breathing, rising vapour and slight camera motion. Fast fragments have directionally correct motion blur without smearing the whole image. No slow motion, speed ramps, freeze frames, HUD or flashy transitions. No cartoons, cel shading, game footage, glossy toys, sculpture showcases or obvious CG-cinematic finish. CITY, WALL AND SCALE A dense medieval stone-walled city: red-brown tiled roofs, pale grey/cream plaster, dark-brown timber framing, narrow paved streets, chimneys, bell towers and pointed spires. Roofs recede in layers; another defensive-wall silhouette appears in the distance. Outside are low woodland, grass and gently receding terrain. A continuous monumental grey-white wall extends into the distance with repeated vertical buttresses and small repeated masonry details along its top. It towers over ordinary houses. The gate sits at the wall's base, with an interior masonry gatehouse, dark archway, supporting pillars and adjacent half-timbered houses. Keep gate, streets, towers, roofs and the dust-wave route geographically consistent between cuts. Only ONE creature exists. It remains OUTSIDE the wall, facing the city, with both hands braced on the wall top and its head and upper shoulders visible above it. Ordinary three-to-five-storey houses are tiny against it, only around shin height. Never place it in an interior street, shrink it to a tall human, move it through the wall or make it climb over. Foreground occlusion must not merge it with buildings. CREATURE IDENTITY AND ACTION LOCK Use the same original creature from @img1 throughout. Tailless colossal biped with a slight forward lean, short thick neck, broad high shoulders, deep barrel torso, long heavy arms, powerful short thick legs and broad feet. Broad low wedge-shaped head, short blunt muzzle, two small nostril slits and a wide CLOSED mouth. No human nose bridge, external ears or exposed teeth. Small deep-set dark amber eyes do not glow. Muted-ochre natural keratin plates cover the low brow. Exactly two short flattened backward-swept cranial fins keep their reference shape and direction. Do not substitute a human face or recolour an existing screen character. The entire body is covered by opaque iron-grey pebbled and scaled thick hide, with natural folds, dust and wear. Ochre plates cover brow and shoulders; broad flattened overlapping dorsal scutes form a central series down the upper back, with smaller matching shoulder scales. These grow organically from the skin, not worn armour, metal or assembled stone. Each hand has FOUR thick digits including an opposable thumb. Each broad foot has THREE short blunt toes and an intact dark-grey leathery padded sole. No red fibres or human-shaped sole. All joints remain naturally connected. No tail, hair, clothing, weapons, machinery, glowing cracks, exposed muscle, blood, wounds or internal anatomy. Thick white-grey vapour rises and curls from close to the body, with believable wind response, volume and occlusion. It is not fire, fireworks or a transformation. It may partly obscure the animal but must not hide its defining silhouette throughout. ONE left-leg windup leads to ONE corresponding gate kick. Left and right are the CREATURE'S own sides: RIGHT leg planted and weight-bearing; LEFT knee bends and LEFT foot draws backward. From directly behind, the lifted leg is screen-left and the support leg screen-right. Both hands remain braced on the wall, torso leans forward, and shoulders/right leg take the weight. The left leg stays attached to the left hip. No mirrored swaps, extra limbs or sole turning into a second face. Successive camera angles show continuous stages of the SAME windup, never a reset. Drawing the foot back prepares a forward kick at the lower gate, not backward walking, stepping over or climbing the wall. Actual contact occurs offscreen and is revealed by the gate bursting inward. No foot-through-wall insert, second kick, punch or body charge. PHOTOGPT ENGRAVING LOCK The only branding is on the central INWARD-FACING wall surface beneath the creature, matching @img2. The complete single-line horizontal wordmark follows @img3: Ph + aperture symbol replacing the FIRST o + toGPT. The second o remains an ordinary letter. Preserve aperture blades, letter order, case, typography and proportions. No separate icon above text, duplicated icon or misspelling such as Photogopt. Carve the wordmark directly INTO the existing pale masonry. Side-lit edges and recessed shadows reveal its depth; retain stone grain, shallow weathering, small chips and dust. No paint, metal plaque, rectangular billboard, coloured background, glow or floating decal. The engraving exists from the start and stays fixed in world space; it never drifts, warps, flips, duplicates or suddenly appears. Only camera angles facing that inner wall can see it. At distance it scales naturally; towers, roofs or vapour may physically obscure it. Do not copy it onto the outside of the wall. The 8.40–11.10-second tower-framed view is the primary readable window. Keep the shot natural: no advertising close-up, freeze or logo automatically facing the lens. The single breach affects the lower gate, while the upper engraved wall section and remaining wall stay standing, though later dust may obscure them. Other stone plaques are blank. No flags, heraldry or decorated banners. Do not put branding on civilians, the creature, rooftops or flying fragments. No other readable text, website, QR code, SIGN UP FREE, slogans, watermark or promotional end card. CIVILIANS — TEXT ONLY All civilians are fictional adults in plain medieval cloth clothing, waistcoats, long skirts, aprons and caps, never modern dress. Photographed-extra realism: natural skin, individual hair strands, believable expressions, fabric weight and continuous gait. No game NPC or wax-dummy appearance. Woman: the SAME blonde adult woman in a light long-sleeved blouse, light apron and red-brown/dark burgundy long skirt resists the wind and is lifted away. Preserve face, hairstyle and clothes between her actions. Final foreground man: adult with short light-brown hair, beige shirt, dark-green sleeveless waistcoat and brown trousers, running toward the camera. Others flee independently rather than forming orderly lines or duplicating one motion. EXACT TIMELINE — 13 SHOTS, 24.00 SECONDS 0.00–1.60 — AERIAL CITY AND SHADOW Open immediately on an extremely high downward view inside the city. Dense red roofs and pale streets fill the frame. A gigantic shadow with broad head, high shoulders and long arms covers the central neighbourhood; pedestrians and ordinary roofs appear tiny. Ease the camera downward with restrained real three-dimensional parallax, preserving the proportions of houses, alleys, people and shadow. Do not begin on a frontal wall image or show the reference board, text or interface. 1.60–4.00 — WALL-TOP FACE AND HAND Hard cut to a slightly canted low-angle close-up from inside the city, looking toward the wall top. Match the face to the board's head-and-shoulders panel. The huge head looks downward; a dark wall edge cuts diagonally across the lower foreground, and a large hand on screen-right grips the stone. Let near-hand perspective establish scale. The head subtly straightens while the eyes remain focused below. Brow growth ridges, small scales, nostrils and eye sockets become clearer. No speech or added open-mouth roar. Warm rim light, rising head/neck vapour and a slow camera approach. Fingers stay in contact with stone. The logo need not be visible from this angle; do not move it. 4.00–5.60 — EXTREME CITY/WALL WIDE Cut to a very wide rooftop-height view. Red tiles, timber gables and chimneys fill the foreground, with the complete tall wall across the distance. Only this creature's head and hands appear at one central location above it. Almost locked-off camera with the slightest lateral drift; emphasize scale through the number of houses and wall length. The existing engraving may appear at a naturally small distance scale, never as an overlay. Wall remains intact; no premature explosion or second creature. 5.60–8.40 — REAR THREE-QUARTER, LEG LIFT AND SOLE Cut outside to a high, slightly canted rear three-quarter view. Creature faces away from the camera toward the city, both hands braced on the wall, RIGHT leg planted. The LEFT knee bends and LEFT foot lifts backward. Follow slightly downward as the huge left sole enters from the bottom into extreme near foreground and progressively occupies much of the frame. Retain torso, arms, wall and roofs as spatial references. Match dorsal scutes and limb proportions to the board's back view; keep three blunt toes, thick dark-grey leathery sole and correct joint connections. The sole belongs to the same attached foot, never another face. Vapour curls upward around the leg. Slow, heavy effort. This camera faces the OUTSIDE wall, so no inner-wall logo is visible. 8.40–11.10 — TOWER-FRAMED BRAND VIEW Cut inside to the wide architectural composition of @img2: large domed bell tower close on the left, tall slender pointed tower close on the right, layered roofs below, and the monumental wall between them. The distant creature's head and shoulders occupy only a small upper-central part of the image; never enlarge its forehead to fill the background. Preserve the layered space and warm golden light. Its head rises slightly and looks into the city while both hands stay on the wall. Foreground towers remain stationary rather than following the head. Beneath the creature, the SAME central PhotoGPT horizontal stone engraving is clearly readable as real recessed masonry. Use only a gentle push, allowing at least approximately TWO seconds of stable legibility while vapour, airborne particles and subtle body motion continue. The left leg stays raised offscreen; do not restart its windup. 11.10–13.10 — GRASS-LEVEL REAR, JUST BEFORE THE KICK Cut outside to an extremely wide rear view from near the grass. Defocused foreground blades sway and produce parallax; nearby trees are far shorter than the creature. Maintain the reference back anatomy but show the braced action pose. Raised LEFT leg is screen-left; planted RIGHT leg is screen-right. Hands remain on the wall and steam rises from shoulders and legs. Windup reaches its peak, then the left leg starts driving toward the low gate. CUT BEFORE CONTACT. No second leg lift or premature breach. 13.10–14.40 — INTACT GATE, THEN THE SINGLE BREACH Hard cut to ground level INSIDE the gate. Huge buttresses and a dark arch extend overhead; a half-timbered house presses against the right frame edge; three ordinary civilians occupy the lower foreground. Hold the intact gatehouse for approximately 0.35 seconds of tense waiting. The nearest civilian notices something and abruptly raises an arm or turns. Then the SAME offscreen LEFT-foot kick bursts the lower gate masonry inward. Backlight penetrates the breach; stones, timber and dust travel along the street toward the camera. One brief violent camera jolt at the rupture, then retain readable spatial direction. Destroy only this gate area, not the entire wall. No foot-through-wall insert or creature entering the city. 14.40–15.60 — GROUND-LEVEL DEBRIS Cut low to follow flying rubble. Unequal stones move rapidly away from the gate, rotate independently and fall under gravity. Near fragments have strong directional blur; a few strike the paving and bounce. Grey-brown dust races along the stone street as the camera briefly rolls. Nothing floats, freezes or glows. All movement comes from the same breach; no new explosion. 15.60–17.30 — WOMAN BRACES, THEN IS LIFTED AWAY Cut to a slightly canted medium-wide street-side view. The blonde adult woman stands before a half-timbered house. Wind pressure travels screen-left to screen-right. FIRST she bends her knees, hunches her shoulders, raises both arms to shield her face and struggles to stay planted. Hair, pale apron, red-brown skirt, nearby tree canopy and loose fragments all stream to the right. THEN stronger pressure unbalances her. Arms leave their defensive position, BOTH feet leave the ground, and she is carried to the right and out of frame. Preserve the order: brace, lose balance, lift, exit. No voluntary jump, ordinary running, landing shot or visible injury. 17.30–18.10 — LOCAL ROOFLINE DAMAGE Cut upward along upper storeys of the narrow street. The SAME shock and debris sweep across eaves and walls. Roof tiles, timber strips and pale plaster fragments detach in sequence; local sections crack and dust curls into the street. Briefly follow the destruction direction. Sunlight still enters between roofs. No spreading inferno or second explosion. 18.10–19.10 — NEAR BLOCK AND DUST CLOUD A quick close insert: a heavy grey architectural fragment rotates past the camera, smaller pieces tumble at different depths, and grey-brown dust surges forward, briefly obscuring buildings. Preserve solid masonry thickness, fracture surfaces, weight and locally readable edges. No glowing orb, meteor or abstract energy effect. 19.10–21.20 — AERIAL PROPAGATING DUST WAVE Cut to a canted high aerial above continuous red roofs, towers and streets. Pale grey dust propagates from the SAME gate breach down streets and over roofs. Large fragments pass close to the camera, with near/far scale creating strong depth. Track forward along their propagation direction; a diagonal dust band crosses the city while roofs and roads remain intermittently visible. Distant areas respond later rather than exploding simultaneously. The rest of the wall remains upright and the creature stays outside. 21.20–24.00 — FLEEING MAN, HOLLOW RECTANGULAR MASONRY LANDING Cut to a low upward street view. Two rows of half-timbered houses form a deep corridor; distant pale green-grey twin towers remain visible, and warm grey dust covers the sky. The short light-brown-haired man in beige shirt, dark-green waistcoat and brown trousers runs continuously toward the camera. Other adults flee to the sides or forward, with continuous strides, arm swings and displacement. Nobody stands watching or lines up to make an orderly path. A huge HOLLOW architectural masonry fragment approaches from high in the distance. It is a thick, broken, approximately RECTANGULAR wall section with a clearly SQUARE OR RECTANGULAR through-opening, substantial fractured edges and a dark interior cavity. Maintain this angular hollow shape. Do not replace it with a circular hole, rounded arch, solid sphere, smooth boulder, meteor, complete house or vehicle. It rotates continuously while descending and rapidly grows in apparent size above the street. Follow its fall with a fast downward and slightly sideways camera movement. FIRST GROUND CONTACT occurs around 23.20–23.50 seconds. The street edge and erupting dust partly obscure the contact point, after which its loss of balance, pronounced sideways tilt and TOPPLING must be clearly visible. Chips and dust spread along the ground. Leave perceptible time after contact for the topple. End at exactly 24.00 seconds with the fragment already grounded and actively toppling, while debris and dust are still moving. Do not end with it suspended in the air. No still frame, black screen, fade or end card instead of landing. No close-up of anyone being struck. MOTION AND CONTINUITY Complete all 13 shots in the stated order over exactly 24.00 seconds. First half: oppressive observation, enormous scale and slow weighty preparation. Second half: one gate breach drives rubble, wind pressure, civilian reactions, local architectural damage, city dust and the final landing. Compress through shot duration, not global acceleration; never omit ground contact and toppling. One creature, one LEFT-leg windup, one gate kick, one persistent breach. RIGHT leg supports and LEFT attacks from every angle; no mirrored swaps. Loaded fingers and hands remain in contact with the wall, without sliding, hovering or fused limbs. Body vapour rises; demolition dust spreads from damaged structures. Distinguish both sources. Individual stones rotate and fall, then create local dust when they contact the ground. Hair, clothes, trees and small fragments respond to consistent wind direction. No extra creatures, modern objects, unexplained transformation, blood, wounds, exposed organs or dismemberment. ORIGINAL CINEMATIC MUSIC AND SOUND Use an ORIGINAL instrumental dark epic-fantasy score: low strings, gradually tightening string textures, restrained weighty brass, deep drums and orchestral percussion. Do not reproduce an existing song or identifiable film/anime theme. No singing, choir, lyrics, dialogue, narration or intelligible shouted words. 0.00–4.00: low strings and sparse dark tones establish unease, leaving space for wind, steam and small falling stone chips. 4.00–11.10: slowly build strings and low brass through scale views, leg lift and tower framing. No brand jingle or jaunty rhythm. 11.10–13.10: tighten tension, then briefly lower the music approaching the gate cut while environmental sound continues, leaving dynamic headroom for the breach. 13.10–14.40: preserve the short intact-gate wait; ONE weighty orchestral accent aligns with the actual masonry rupture. Physical impact and fracture transients remain prominent. No premature hit implying another collision. 14.40–21.20: urgent strings and measured low percussion sustain the destruction without restarting at each cut or quantising every stone impact to a beat. 21.20–24.00: rise to a final climax during the hollow fragment's descent. Emphasise first ground contact while foregrounding landing, toppling, fracture and rolling sounds. End with the 24-second film, without an additional outro, fade or end card. Synchronise physical sound with action, distance and occlusion: interior/exterior wind, nearby grass friction, body steam, stone chips under gripping fingers, immense weight shifts, masonry strain, the SINGLE gate impact, breaking stone and wood, flying fragments, bounces, wind-tugged fabric, wordless rapid breathing, running footsteps, falling tiles, cracking walls and the final hollow block striking, toppling and scattering chips. Mix with real dynamic range and near/far depth. Music supports rather than continuously obscures physical effects. No constant maximum loudness, clipping, distortion or muddy low-frequency overload. FINAL OUTPUT Exactly 24.00 seconds, landscape 16:9, full-frame. No black bars, rounded borders, titles, captions, playback controls, interface, watermark, promotional outro or frozen final frame. The only readable brand is the fixed PhotoGPT horizontal recessed carving on the inner central wall. Preserve one original creature, one continuous city, coherent lighting, photographic optics and physical motion throughout. End with the landed masonry toppling and dust still spreading.

Reference Images

@itsphotogptai

You may also like