電影感 AI 影片提示詞模板

A tiny dragon visits a castle kitchen. A nervous chef offers dinner, then discovers the visitor prefers dessert. Warm firelight, expressive reactions, a playful cinematic camera move.

Create a 10-second cinematic first-person POV video from inside a refrigerator looking outward, as if the camera is hidden behind cold shelves, glass bottles, and food containers. Keep the fridge frame, bottles, and containers close to the lens as foreground depth. At 0-2s, a sharply dressed man opens the refrigerator door. Cold white fridge light floods hi

Create a 10-second cinematic 2D American comic animation from a first-person POV inside a refrigerator looking outward. The camera is hidden behind cold shelves, milk cartons, glass bottles, and snack boxes, making the viewer feel physically inside the fridge. Use input[0] as Character 1, a stylish teenage boy. Use input[1] as Character 2, a stylish teenage

Create an 8-second cinematic Korean neo-noir apartment suspense video from a true peephole POV, using input[0] as the visual reference for the visitor, wardrobe, hallway style, and color palette. The camera is inside an apartment door looking outward through a round door peephole. Keep a subtle circular peephole feeling throughout: soft dark round vignette a

Create a 7-second cinematic black-comedy crime video from a true trash can POV in 16:9 widescreen, using input[0] as the visual reference for character design, wardrobe, props, alley mood, and color palette. The camera is fixed near the bottom inside a street trash can, looking upward through the open rim. Keep the dirty trash can rim and pieces of garbage c

Cinematic live-action video, about 10 seconds. Static POV from inside the open trunk of a car, looking upward and outward through the trunk opening. The trunk interior creates a dark curved frame around the image, with the rubber seal and side panels visible at the edges. Nighttime roadside police inspection, red and blue patrol lights flashing softly in the

Cinematic live-action ocean thriller video, 16:9 landscape, about 10 seconds. Underwater predator POV from below the ocean surface, looking upward toward the bright moving waterline. The camera represents a shark-like predator moving through the water, but the animal itself is never visible. The surface above shimmers with sunlight, waves, bubbles, and suspe

2D animated dark fairytale horror video, 16:9 landscape, about 8-10 seconds. Static POV from underneath a bed, looking outward from the floor-level space beneath the bed. The camera remains fixed under the bed for the entire shot. The left and right sides of the frame show bed legs and the lower edges of the bed frame. The edge of the bedsheet hangs down nea

Cinematic live-action crime thriller video, 16:9 landscape, about 10 seconds. Fixed overhead black-and-white security camera POV from the ceiling corner of a small convenience store, looking down at a realistic store layout. The camera remains static for the whole shot. Monochrome surveillance footage, fluorescent store lighting, mild camera noise, crisp ove

Cinematic wilderness thriller video, 16:9 landscape, about 10 seconds. The entire shot is seen through a sniper scope view from a hidden observer on a rocky ridge. Use input[0] as the monster reference. Preserve the monster's main silhouette, head shape, body proportions, skin texture, distinctive ridges or horns, limbs, and eye glow. Visual framing: the vi

Found footage horror video, 16:9 landscape, about 10 seconds. DV camera recording perspective. Use input[0] as the protagonist who appears at the beginning of the recording. Preserve her face, hairstyle, clothing style, and overall identity. The video begins with the protagonist speaking directly to the DV camera in a dark hallway. After she speaks, the DV

Found footage monster video, 9:16 vertical, about 10 seconds. Daytime DV camera recording perspective on a modern city street. Use input[0] as the lizard-man creature. Preserve the creature's lizard head, amber eyes, scaled dark green body, claws, tail, and Japanese tokusatsu rubber-suit texture from input[0]. The camera operator never appears on screen. If

Cinematic grounded superhero scene, realistic Korean cinema, 16:9, one continuous wide shot. Use input[0] as the main character identity: an adult young man with the same face, hairstyle, outfit, and calm confident charisma. Composition and lighting are critical: place the character on the right-side two-thirds area of the frame, not centered. Keep him seve

Cinematic daytime apartment-window observation shot, classic voyeuristic courtyard perspective, 9:16 vertical video, one continuous slow camera movement. Camera and framing: The camera is inside one apartment, observing the opposite apartment building across a narrow courtyard through a window or balcony frame. It feels like a quiet observer watching neighb

Use input[0] as the main character appearance reference. Keep the rider's identity consistent: same face, hairstyle, skin tone, outfit colors, body type, and silhouette. Create one continuous 12-second original cinematic fantasy aerial action video in 16:9 with English sound. The referenced student rides a massive original dark winged dragon-like creature t

Use input[0] as the main character appearance reference. Keep the rider's identity consistent throughout the video: same face, dreadlocks, skin tone, outfit colors, body type, and silhouette. Create one continuous 12-second original cinematic jungle raptor-riding adventure video in 16:9 with natural sound. The referenced woman rides a large agile feathered

[Quality and Technical Specs] Generate a 15s, 16:9 AI video template titled "Snow Fortress Archer Rescue". Use snowy fortress rescue drama with heroic archery, slow-motion arrow tracking, and cinematic wind movement. The output must be polished, stable, social-video-ready, and visually coherent. [Hard Settings] input[0] is the sole visual identity reference

[Quality and Technical Specs] Generate a 10s, 16:9 text_to_video AI video template for Stealth Bomber Snow Strike. Style: military aviation action with low-altitude stealth penetration and explosive mountain impact. The video must be polished, stable, rhythm-driven, visually coherent, and suitable for reusable template generation. [Hard Settings] No input r

[Quality and Technical Specs] Generate a 10s, 9:16 refs_generate AI video template for Jdrama Romance Moments. Style: cinematic reusable social-video template. The video must be polished, stable, rhythm-driven, visually coherent, and suitable for reusable template generation. [Hard Settings] input[0] is the character reference image. Strictly use input[0] a

[Quality and Technical Specs] Generate a 15s, 9:16 text_to_video AI video template for Tennis Ball Flight POV. Style: premium sports commercial with bright Paris travel energy. The video must be polished, stable, rhythm-driven, visually coherent, and suitable for reusable template generation. [Hard Settings] No input reference image is required. Generate th

[Quality and Technical Specs] Generate a 15s, 9:16 refs_generate AI video template for Female Warrior Battlefield Charge. Style: epic battlefield action with horse charge and practical combat sound design. The video must be polished, stable, rhythm-driven, visually coherent, and suitable for reusable template generation. [Hard Settings] input[0] is the char

input[0] is the character reference image. Strictly use input[0] as the visual identity reference for the main character. Below-frame entrance + orbiting camera. Korean cinema texture: a dim VIP room with dark red velvet walls, a swaying tungsten chandelier casting mottled shadows overhead, poker cards and whiskey glasses scattered on the table. Teal-orange

Here's the English translation of your prompt: --- "input[0] is the character reference image. Strictly use input[0] as the primary visual identity reference for the protagonist: extract and maintain the same person's face, skin tone, hairstyle, body type, costume, accessories, aura, and overall style. Do not show the reference image itself. Do not generat