carat

Prompt Gallery

/

Surreal

Login
Sign Up

Surreal

AI video prompts for dreamlike surreal scenes. Quickly create distorted spaces, floating objects, and dreamscape cuts.

2026.05.22

·
AI-generated Element Emerging From Polaroid made with Nano Banana Pro

Element Emerging From Polaroid

Nano Banana Pro· Image
A hyper-realistic 3D photo concept showing a hand holding a square polaroid photo labeled “ELEMENT”. Inside the photo, there’s a close-up of [ELEMENT], and it flows seamlessly out of the photo frame into the real world, naturally interacting with the surface below. The text “ELEMENT” appears printed cleanly on the white border of the photo. Realistic cinematic lighting, ultra-detailed textures, artistic composition, 1080x1080 square format.
@TechieBySA

Magical Autonomous Painting Atelier

Seedance 2.0· Video
[CONDITION DEFINITION] Create a 15-second cinematic time-lapse video inside a warm atmospheric artist’s atelier. A blank white canvas stands on an easel from the beginning. Use the uploaded 1:1 reference image only as the subject of the final painted artwork on the canvas. Do not show the uploaded reference image itself at any time. Do not place its subject in the room as a real person, background figure, poster, printed image, overlay, or pre-existing artwork. The subject from the uploaded reference image must exist only as the final painting on the canvas. Brushes, palette knives, and the paint palette move autonomously by themselves. Colorful paint rises from the palette and dances through the air like smoke, ribbons, thread-like streams, and living brushstrokes. The final result should become a rich oil-painting interpretation of the uploaded reference image. [OPTIONAL PERFORMER LOOK] No performer. No painter. No human hands. No human arms. No visible human body. All tools move autonomously. [SHOT / FLOW] 0-2s: Establish the atelier. Blank white canvas on an easel, palette, brushes, paint tubes, warm studio light, subtle creative mess. The canvas is completely empty. 2-5s: The tools begin moving by themselves. Paint slowly lifts from the palette and starts dancing in the air. 5-9s: Fast magical paint choreography. Color trails swirl, weave, and spin through the studio like elegant smoke and ribbons. Use rapid jump cuts. The canvas remains mostly abstract, with only partial marks, incomplete colors, and no recognizable subject. 9-12s: All paint trails suddenly rush toward the canvas. In a rapid montage, sketch marks, bold color blocks, texture, highlights, and key shapes assemble quickly. Keep the image mostly abstract and incomplete until the end of this section. Do not reveal the recognizable subject before 12 seconds. 12-15s: Reveal the finished painting clearly for the first time. Show a satisfying close-up of the completed artwork, clearly based on the uploaded reference image, then a brief lingering afterglow in the atelier. [CAMERA / EDITING] Dynamic and energetic. Use quick push-ins, slight pull-backs, macro close-ups of wet paint and brush bristles, medium shots of the easel, overhead angles, side angles, and rapid jump cuts every 0.5 to 1 second. The middle section should feel like a fast magical montage, not slow continuous motion. Keep the canvas mostly abstract until 12 seconds. The studio is only the setting; the subject from the uploaded reference image must never appear outside the canvas. [SOUND] No background music. Sound effects only. Use satisfying brush strokes, wet paint movement, palette knife scraping, paint tube squeezes, soft whooshes, and subtle studio ambience. [NEGATIVE] Do not show the uploaded reference image as a separate photo, screen, comparison, poster, printed board, overlay, or background image. Do not place the uploaded subject anywhere in the studio outside the canvas. No real person version of the uploaded subject. No human hands, no human arms, no painter, no performer focus, no multiple people, no cartoon character, no mascot, no text, no captions, no numbers, no timestamp text on screen, no logo, no watermark, no signature, no split screen, no poster layout, no static single-angle shot, no early full reveal of the final painting, no unfinished ending.39
YouMind

Infinite Library Black Hole Sci-Fi

Seedance 2.0· Video
15-second ultra-cinematic sci-fi fantasy sequence inside an infinite library orbiting a gigantic black hole in deep space. Endless bookshelves stretch in impossible directions, curving upward and sideways under distorted gravity. Ancient books, scrolls, glowing manuscripts, and floating pages drift weightlessly between massive wooden staircases and suspended reading balconies. The camera glides smoothly through the library as the architecture slowly rotates around the black hole visible through enormous shattered windows. The black hole bends starlight into a glowing accretion ring of gold, blue, and violet energy. Gravity behaves strangely: books float upward, chandeliers hang sideways, dust particles spiral in slow orbit, and ink drips through the air like liquid stars. A lone librarian figure walks calmly along a floating staircase, carrying a glowing book while pages orbit around them like tiny moons. Mid-sequence, the black hole pulses. Shelves bend slightly, time ripples through the room, and thousands of books open by themselves, releasing luminous symbols into the air. Final shot: the camera pulls back to reveal the entire infinite library wrapped around the black hole like a colossal ring of knowledge, glowing softly in cosmic darkness. Style: ultra-cinematic cosmic fantasy, surreal gravity physics, impossible architecture, volumetric starlight, floating architecture, volumetric starlight, floating particles, glowing manuscripts, black hole lensing, majestic scale, no text, no overlays. Audio: deep cosmic ambience, low gravitational rumble, pages fluttering, wooden shelves creaking, distant celestial resonance, soft magical energy hum.
YouMind

Palace Traditional Dance Fantasy

Seedance 2.0· Video
Absolutely no subtitles, no text overlay, no watermarks, no title cards, no credits, no on-screen text of any kind. 16:9, 15-second seamless 3-shot progression. Inside a vast abandoned Korean palace hall flooded with two inches of mirror-still water at blue hour. Ornate wooden columns with faded dancheong patterns rise into darkness. A lone Korean woman in her late twenties wearing a flowing white hanbok with long silk sleeves stands at center stage. Shot 1 — extreme macro from beneath the water surface as her bare foot touches down, sending silver droplets and concentric ripples outward. Realistic water refraction. Silk sleeve tips drag through the surface. Her visible breath curls in the cold air. [HARD CUT TO — smooth 180-degree orbit, waist level] She begins a traditional salpuri dance spin — each footstep ignites pale blue bioluminescent constellations beneath the water. Reflections of the ornate ceiling ripple and distort. Ancient brass lanterns along the columns flicker awake one by one. Dust particles drift through volumetric shafts of moonlight. [HARD CUT TO — rapid pull-back, rising crane] On the final accelerating spin, the brass lanterns burst into hundreds of luminous white paper cranes spiraling upward through the broken roof. The camera rockets backward and up to reveal the entire palace hall floating alone at the center of a dark mirror-ocean at sunrise. Golden rays pierce through parting clouds. Her tiny white silhouette still spins at the glowing center far below. Audio: begin with dripping water echoing in stone, a faint inhale. Precise water splashes synced to each step. A low gayageum string swell building with the spin. Paper wing flutter overhead. Final reverb dissolving into ocean silence.

Stone Colossus Flower Field

Seedance 2.0· Video
Absolutely no subtitles, no text overlay, no watermarks, no on-screen text of any kind. A colossal ancient stone colossus — 200 meters tall, covered in moss and weathered cracks — walks slowly through a vast field of yellow canola flowers under bright golden afternoon sun. Jeju Island Hallasan mountain visible in the distance. Each massive stone footstep lands gently. Instead of destruction, the impact sends a shockwave of blooming — thousands of new flowers explode outward from each footprint in expanding rings of yellow, orange, pink, and white. Butterflies and petals launch skyward with each step. The colossus moves with gentle slow purpose, tiny birds nesting in its stone shoulders. Camera tracks from a low angle among the flowers looking up at the giant silhouetted against blue sky and golden sun. Audio: deep earth-shaking thuds softened by flower fields, explosive blooming whoosh, birdsong, wind through petals. No music.
AI-generated Dusk Shindig with Multiple Fake Imagination Projections All Aligned in the 14th Dimensions made with GPT Image 2

Dusk Shindig with Multiple Fake Imagination Projections All Aligned in the 14th Dimensions

GPT Image 2· Image
A dusk shindig with multiple fake imagination projections all aligned in the 14th dimensions
@workingclassbud
AI-generated Soft Sculpture Jumping Character made with GPT Image 2

Soft Sculpture Jumping Character

GPT Image 2· Image
A complete single 3D illustration, exaggerated artistic editorial illustration style, not storyboard, not a collage layout. The subject is an exaggerated cartoon character with a tiny head, round chubby body, extremely elongated limbs, oversized hands and oversized shoes, off-balance center of gravity, frozen mid-air in a dramatic jumping pose, with strong tension and comedic energy. Overall form looks like a soft sculpture toy round, chunky, bouncy, exaggerated not anatomically correct. Materials are matte soft rubber, fuzzy fabric, knitted texture, plasticine feel, with fine fiber details, slight grain, hand-sculpted touch avoid glossy plastic toy look, avoid translucent glass look, avoid high reflections. High-saturation dopamine color palette, bold color clashes, large flat color blocks, bright but not over-glowing. Clean white background, minimal space, only a soft oval shadow on the ground, no complex scene. Floating abstract graphic elements around the character: stars, squiggle lines, balls, cubes, icons, symbols all rendered as soft rubber or paper props to enhance motion and design feel. Soft studio lighting, global illumination, soft shadows, low contrast, clean commercial render. C4D / Blender 3D illustration, stylized soft sculpture, matte clay material, fuzzy fabric texture, knitted surface detail, playful editorial 3D illustration, high quality. Theme: a SaaS-hoodie product designer floating with a laptop, surrounded by dashboard UI cards, charts and task widgets, hero illustration for a SaaS landing page
@MissDelulu9
AI-generated Floating Streetwear Fashion Poster made with GPT Image 2

Floating Streetwear Fashion Poster

GPT Image 2· Image
一张高端潮流街头服装海报,画面主体是一位无脸的未来感模特漂浮跳跃在天空中,穿着超大号米色羽绒服、层叠兜帽和针织面罩,面部被完全遮住;下身穿白色宽松工装裤,裤腿外叠加夸张的浅蓝与深蓝堆叠牛仔护腿,脚穿米色运动鞋,背着黑色单肩包。 主体采用极低机位仰拍视角,腿部和鞋子被夸张放大,形成强烈的视觉冲击,仿佛人物正在从观众头顶跃起。动作充满反重力感,一只手向外伸展,一条腿向前踢出,整体姿态动感、飘浮、超现实。 背景是明亮的青蓝色天空和柔软白云,日光清透,空气感强。整体融合时尚大片摄影、超现实潮流广告和街头海报设计。在画面右上角和左下角加入两个缩小版的人物剪影拼贴,一个在空中漂浮,一个正在跳跃,形成时装杂志拼贴感。 海报排版为方形杂志封面设计,顶部加入巨大的抽象圆角字体,类似 Y2K 泡泡字和实验性街头品牌 LOGO;左侧加入小型涂鸦风品牌标识;右侧加入白色手写刷字标题“RISE ABOVE WITH FASHION”,下方配极小号编辑说明文字;底部中央加入小型条形码。整体带细白边框和淡米白纸张纹理。 风格关键词:未来街头时尚、先锋服装设计、Y2K 平面设计、杂志封面排版、拼贴剪影、超低机位仰拍、天空背景、超现实漂浮、夸张透视、潮流广告大片、真实布料质感、高清细节、干净高级。 Make the aspect ratio 9:16
@ponyodong

Brand Concept Film Through Windows and Eyes

Seedance 2.5· Video
A cinematic brand concept short. @Image1 is the first frame; the picture shakes slightly, the camera gradually pushes in, arriving at tree shadows rushing backward outside a window, the shadows receding faster and faster, then suddenly cuts to @Image2 where the speed abruptly slows; the camera moves gently forward along a stream, birdsong and flowers. The camera moves down, arriving underwater, with the sound of bubbles (@Image3 @Image4); a group of orange jellyfish drift gracefully across the frame. The camera slowly pulls back; a school of small fish flit across and pass from the water into the window. A girl looks left and right, watching the little fish. The camera slowly pulls back, the image goes out of focus, then refocuses and becomes clear, switching in time with the music: a Chinese-garden lattice window (@Image5) with light turning, a church stained-glass window, an airplane cabin window, a domed skylight, a black maple window, a louvered blind, a European dormer window, a peephole in a door, a camera viewfinder frame, a bird's eye, a human-eye close-up. The frame rests on the human-eye close-up, then the eye closes, the screen goes black, then suddenly the eye opens and the word 'seedance' appears in the center of the eye with an accent.
@johnAGI168

Dynamic Time-Lapse with Static Subject

Seedance 2.0· Video
keep the central person completely motionless and frozen like a statue. Animate only the surrounding background and people with fast-paced movement and heavy motion blur, creating a dynamic time-lapse effect around the static subject. The camera must remain perfectly fixed.
YouMind

Found Footage Graveyard Creature Horror

Seedance 2.0· Video
Low-light handheld smartphone footage, heavy grain, slight motion blur. Harsh LED phone flashlight only. Natural audio: insects, distant dogs, wind moving dry leaves. Location: an old Indian graveyard with uneven mud paths, cracked tombstones, iron fencing, overgrown grass, and leaning graves. On top of a large, cracked tombstone, a woman is crouched unnaturally. Unnaturally thin body, grayish decaying skin with two heads clearly visible. Long tangled black hair hanging over her face. Milky white unfocused eyes. Mouth slightly open, revealing sharp Dracula-like fangs. A low guttural growl rises from her chest, mixing with the night sounds. Her head turns slowly toward the camera, hair sliding aside just enough to reveal her dead white eyes. Sudden violent movement as she launches off the tombstone, landing with a wet thud. Panicked retreat between graves. Camera jerks violently. She drags herself across the ground and over headstones, limbs bending at impossible angles, snarling, dirt and dry grass scraping beneath her movements. Abrupt extreme close-up. Her face slams into frame mid-screech, fangs fully exposed, eyes blank and lifeless, hair whipping forward. Instant cut to black.
YouMind

Tense Countdown Sci-Fi Thriller

Seedance 2.0· Video
Man alone in a dim apartment bathroom at night. Cold fluorescent lighting flickering overhead. A glowing red digital countdown appears beneath the skin near his temple, visible through translucent veins like a holographic neurological implant. Timer starts at 00:10. Extreme close-up on terrified eyes. Heavy breathing. Sweat forming instantly. He touches the glowing area in confusion and panic. 00:9 The countdown pulses brighter. He splashes water on his face aggressively. Mirror reflections distort subtly. He tries scraping at the glowing numbers with trembling fingers. Breathing becomes unstable. Heartbeat-like light pulses spread across his skull. Camera shakes slightly. Bathroom lights flicker harder. He searches frantically through drawers for something to stop it. Close-up of sweat dripping into sink. He presses his head against the mirror trying to crush the countdown visually. Red light leaks through his skin. 00:05 The room hum begins vibrating unnaturally. Audio distortion sensation. His hands shake violently. He takes a towel, desperately. Countdown still visible glowing through fabric. The fluorescent light explodes into darkness. Only the red countdown illuminates his face. Complete panic. Fast breathing. Handheld camera very close to him. Absolute silence. The countdown freezes. The red glow slowly fades away. Power returns softly. He stares at himself in the mirror, shaking, unable to believe he survived. Ultra cinematic psychological sci-fi thriller, realistic acting, grounded visuals, intense tension, sweat, panic, shallow depth of field, realistic skin detail, subtle environmental distortion, dramatic lighting.
YouMind

What Are Surreal AI Prompts?

Surreal prompts are AI video prompts for quickly creating dreamlike scenes that defy logic, cuts where space distorts or objects float, and subconscious-like landscapes. In Carat, you generate scenes with video models like Seedance 2.0, then keep going with follow-up work such as editing the result, changing the aspect ratio, adding captions, or stitching multiple cuts together.

Where You'll Use It

  • Objects and figures floating against gravity
  • Distorted landscapes where space bends or stretches endlessly
  • Dreamy transition scenes where reality and dream overlap
  • Surreal mood cuts with unreal color and inverted scale
  • Extending a concept image into a moving surreal video

Why This Tag Is Useful

Surreal video lives in the balance between breaking the rules of reality and still feeling somehow plausible. More than merely bizarre imagery, it's the dreamlike mood, the soft transitions, and the scale and color that subvert expectation that pull a viewer in. When choosing a prompt, it helps to check not just the subject but also the method of transformation, the speed of transition, color and texture, and the channel you'll use it on.

Tips for Building Your Prompt

  1. Decide the video's purpose and platform first.
  2. Spell out the camera angle, movement, lighting, and background in detail.
  3. If a person or product appears, specify wardrobe, expression, props, and brand elements separately.
  4. After generating, review for explicitness, copyright, logos, likeness rights, and text errors.

In Carat, you can build follow-up clips from a generated video, adjust it to short-form ratios, and add captions, music, and overlays to polish it into content ready to post.

Frequently Asked Questions

Related Prompts

Poster

Poster

574 prompts

Illustration

Illustration

774 prompts

Character

Character

343 prompts

Surreal Video

46 prompts

Carat Prompt Gallery

Create trending AI prompts instantly

Prompts

  • Logo
  • PPT Design
  • Thumbnail
  • Poster
  • Web Novel Cover
  • ID Photo
  • Caricature
  • Product Detail Page
  • Banner

Models

  • GPT Image 2
  • Seedance 2.0
  • NanoBanana Pro
  • NanoBanana 2

Carat

  • Get started
  • About
  • Blog
  • Careers
  • Contact
  • Partnership
  • Terms
  • Privacy

© 2026 Paradot.Inc All rights reserved.