video prompt:
Create an 18-second horizontal 16:9 video using our latest Tom and Jerry desktop wallpaper as the exact first frame. Preserve the warm living room, three-column icon grid, top menu bar, and centered bottom Dock. Replace the frying pan in Tom’s hand with one flat cooking spatula. Jerry carries nothing.
Camera: One continuous static shot. No cuts or camera movement.
0–3s: Jerry runs from right to left. Tom chases him and throws the spatula directly at Jerry. Show the spatula clearly leaving Tom’s paw and flying left in a straight, readable path.
3–5s: Jerry sees it coming and leaps upward. The spatula passes underneath him, misses, and flies straight into the third column of desktop icons. It hits the icons directly—it does not strike or bounce off the floor first.
5–7s: The six third-column icons—Gmail, Discord, Microsoft Teams, Microsoft Excel, Netflix, and Google Drive—come loose and scatter onto the floor. Their six grid positions are visibly empty. The other icons, menu bar, and centered Dock remain still.
7–11s: Tom stops, realizes what he has done, and rushes to the fallen icons. He picks up all six scattered icons from the floor and gathers them against his chest, holding them securely in both arms. Show the floor become clear as he collects them. Jerry watches nearby.
11–17s: Holding the remaining icons against his chest, Tom takes one icon at a time and places it into its correct empty position, from top to bottom: Gmail, Discord, Microsoft Teams, Microsoft Excel, Netflix, Google Drive. After each placement, that icon leaves his paw and stays in the grid; the number held against his chest decreases by one. Nothing snaps back by itself.
17–18s: Tom checks the restored column and gives Jerry an embarrassed glance. Jerry looks amused. End with all 18 icons in their original positions.
Critical action order: Tom throws the spatula → Jerry jumps clear → spatula directly hits the third column → six icons fall → Tom gathers all six against his chest → Tom replaces them one by one. Do not skip, reverse, or combine these actions. No floor ricochet, frying pan, cheese, popcorn, duplicated icons, changing logos, extra limbs, dialogue, subtitles, or camera movement.
video prompt:
Create a 30-second vertical 9:16 live-action supernatural short film. It should feel like disturbing real footage captured by an ordinary person on a cloudy roadside, not a polished movie trailer or a video game. Natural overcast daylight, damp asphalt, muted colors, imperfect handheld movement, slight motion blur, realistic vehicle weight and road physics. No music and no dialogue.
The setting is one continuous two-lane country road bordered by dense green trees. Keep the geography consistent throughout: a white delivery van ahead of the motorcyclist, a large blue cargo truck approaching in the opposite lane, and a narrow grassy shoulder beside the road. The motorcyclist is an adult man wearing a tan jacket, dark pants, and a black full-face helmet, riding a red sport motorcycle. Keep the same man, motorcycle, van, and truck in every shot.
0–6 seconds — From a close, shaky roadside camera angle, the red motorcycle accelerates to overtake the white van. The movement is casual and believable at first. Tire noise, wind, and engine sound are natural. As the motorcycle crosses into the opposite lane, the blue cargo truck suddenly appears directly ahead, much closer than expected. Its headlights and large front grille fill more and more of the frame. The rider has no room to return to his lane.
6–9 seconds — At the exact instant before impact, time stops abruptly. The motorcycle, rider, truck, van, loose leaves, and a small splash from the road freeze completely in place. The engine and traffic sound cut out in the middle of their motion, leaving a faint low hum. The truck is only a short distance from the motorcycle. Hold this impossible, frightening stillness long enough for the viewer to understand it.
9–16 seconds — The camera alone is free to move. It slowly rotates around the frozen rider and pulls backward, as though separating from him. The movement should feel like an unsteady person turning their head, with a subtle floating quality, not a smooth cinematic drone shot. The camera tilts down. Two pale, faintly translucent hands rise into the foreground from the camera's point of view. The rider's physical body remains frozen on the red motorcycle several feet away. This reveals that the camera is now seeing through the rider's spirit. Keep the effect restrained and photorealistic: no bright glow, particles, or fantasy energy.
16–23 seconds — The spirit camera turns toward the grassy shoulder. A tall, motionless figure in a worn black hooded robe stands among the roadside plants, partly obscured by branches. Its face is mostly darkness, with only a suggestion of a skull-like face when the camera settles. It has been standing there quietly the entire time. The camera hesitates, then takes one small, involuntary step backward. No jump scare yet. The truck, rider, and other traffic remain perfectly frozen behind the camera.
23–27 seconds — The spirit camera turns back toward the road. Time resumes without warning. Traffic sound slams back in. The blue cargo truck strikes the red motorcycle with convincing speed and physical force. Show the motorcycle being thrown and tumbling across the road while the rider disappears out of the immediate frame. Film the impact from the spirit's position a short distance away, like a stunned witness who cannot intervene. Use a brief handheld jolt and realistic debris, with no gore and no slow-motion replay.
27–30 seconds — The camera whips back toward the roadside. The hooded figure is now much closer, standing silently in the foreground. Hold its dark face for one unsettling beat, then cut sharply to black. The traffic sound continues for a fraction of a second after the cut.
The central reveal must be clear: the camera leaves the rider's body during the frozen instant, the translucent hands belong to his spirit, and the hooded figure is waiting for him. Maintain realistic scale, consistent vehicle positions, believable collision physics, and an amateur handheld recording aesthetic. No captions, logos, interface elements, dramatic color grading, exaggerated camera moves, cartoon effects, or visible gore.
video prompt:
Photorealistic vertical smartphone video, 10–12 seconds, filmed by a spectator in a crowd. At a forest overlook, a young adult in a bright yellow jacket stands on a wooden platform high above a large white elastic net stretched between several tall trees. The net is far below the platform, roughly 15 meters down, with a clear view of the long drop. Spectators crowd around the platform, some filming on their phones.
The person jumps. The handheld camera tilts down quickly to follow the full descent, briefly struggling to keep them in frame. They land in the center of the white net, which stretches deeply under their weight, then rebounds and launches them back toward the platform. They catch the edge and climb up as the crowd gasps and cheers.
Keep the height and the distance to the net visually consistent in every shot. One continuous spectator phone recording, natural camera shake, imperfect framing, realistic forest ambience and spontaneous crowd reactions. No cuts, slow motion, cinematic camera moves, captions, or gore.
video prompt:
Create a 15-second stylized 3D animated short with a warm, high-end family-film aesthetic. Soft matte materials, expressive character animation, cinematic lighting, no photorealism, no glossy or oily surfaces.
CHARACTER 1 — ROBOT:
A tiny white household robot with a rounded capsule body, no obvious legs, very small retractable arm nubs, a dark digital face panel, huge warm amber glowing eyes, tiny mouth, ivory-white shell, light gray details, and a small amber indicator light. Cute, innocent, slightly clumsy. Keep the exact same design and proportions throughout.
CHARACTER 2 — HUMAN:
An adult female character with a youthful, gentle animated appearance. Petite build, soft facial features, simple casual clothing, quiet and slightly lonely mood. Natural restrained expressions, no dialogue, consistent appearance throughout.
ENVIRONMENT:
A large, empty rooftop of an ordinary residential apartment building at night. Wide concrete floor, faded parapet walls, old pipes, vents, water tank, AC units, and a simple makeshift robot living corner with a tarp, thin mattress, worn blanket, crude desk, stool, storage boxes, and telescope.
The rooftop is mostly dark. The distant apartment skyline is dim and hazy. Heavy blue-gray clouds cover the sky.
IMPORTANT SKY RULE:
There is NO MOON anywhere in the entire video.
No crescent moon, full moon, half moon, glowing celestial disk, planet, or sun.
At the beginning, the sky contains ZERO visible stars.
Only one weak warm lamp near the robot's shelter provides light.
0–3s:
Wide shot of the empty rooftop.
The tiny robot rolls out carrying a glass jar containing several small glowing golden star-shaped lights.
The jar softly lights its face and white shell.
The robot stops near the rooftop edge, looks up at the completely dark, starless sky, opens the jar, takes out one glowing star, and carefully throws it upward.
The glowing star rises only a little, then falls back and lightly hits the robot on the head.
Small comedic “bonk.”
The robot freezes, blinks once, and looks upward in confused disappointment.
3–6s:
The robot picks up the glowing star and tries again.
This time it steps back and throws much harder.
The camera follows the tiny glowing star upward.
For one moment it seems successful—
then it curves backward and drops directly into the open jar.
Tiny glass clink.
The robot stares into the jar.
Silent beat.
Its shoulders slowly drop and its eyes become slightly sad.
6–9s:
The robot hears a small movement behind it.
Camera pans across the dark rooftop.
An adult woman sits alone near the parapet wall, knees drawn close, quietly looking at the dark city.
She is not crying, only calm and lonely.
The robot looks at her.
Then at the jar.
Only one glowing star remains inside.
It hesitates, then its expression softens.
9–12s:
The robot slowly approaches her.
She notices the warm glow and looks up.
The robot stops beside her and carefully offers the final glowing star with its tiny arm nubs.
Close-up:
golden light reflects softly in her eyes.
She gently takes the star in both hands.
Her expression slowly changes into a small genuine smile.
The robot's eyes immediately brighten.
12–15s:
Close-up of the woman holding the tiny glowing star.
Warm light softly illuminates both characters.
She looks at the robot.
The robot gives a shy happy expression.
Then she notices something extremely small in the sky.
She looks upward.
The robot follows her gaze.
Camera slowly pulls back into a wide shot.
FINAL SKY EVENT:
One single REAL STAR becomes visible high in the dark cloudy sky.
It must look like a real distant star:
a tiny sharp white point of light,
almost pin-sized,
very far away,
with only a subtle natural twinkle.
The star must remain extremely small in frame.
It is NOT a five-point cartoon star.
It is NOT a glowing orb.
It is NOT a large light source.
It must NEVER resemble the moon.
NO circular disk.
NO crescent shape.
NO visible surface.
NO large halo.
NO yellow moonlight.
NO planet-like object.
Only ONE tiny distant white point of starlight.
The surrounding sky stays dark and mostly empty.
No other stars appear.
Final shot:
the woman and tiny robot sit side by side near the rooftop edge, quietly looking at that single distant star.
The robot makes one tiny happy wiggle.
Cut to black.
VISUAL STYLE:
Stylized high-end 3D animated feature-film look, soft rounded shapes, matte surfaces, subtle texture, expressive but restrained faces, warm amber light against cool dark-blue night, slightly worn residential rooftop, no excessive fantasy effects.
CAMERA:
Wide establishing shot, medium character shots, reaction close-ups, gentle tracking, final slow pull-back. Smooth camera movement, no handheld shake, no rapid cuts.
AUDIO:
No dialogue, no lyrics, no copyrighted music.
Use distant city hum, light wind, small robot mechanical sounds, glass clinks, soft metallic taps, and a subtle “bonk.”
Minimal music: playful during the failed throws, soft and emotional near the ending.
NEGATIVE:
ABSOLUTELY NO MOON.
NO crescent moon.
NO full moon.
NO half moon.
NO lunar disk.
NO glowing circle in the sky.
NO planet.
NO sun.
NO large celestial object.
No subtitles, text, logos, watermark, neon city, crowded rooftop, magical explosion, excessive particles, star field, multiple stars, outfit changes, character morphing, glossy plastic skin, excessive bloom, or live-action realism.
A tiny robot keeps trying to return a star to the sky, but ends up giving the last one to someone feeling lonely ✨
This animated short was generated with MiniMax H3 dev.
The price just dropped from 40% → 30%, making it even more affordable to create Pixar-style animated shorts.
Full tutorial below 👇
video prompt:
Create an 18-second horizontal 16:9 video using our latest Tom and Jerry desktop wallpaper as the exact first frame. Preserve the warm living room, three-column icon grid, top menu bar, and centered bottom Dock. Replace the frying pan in Tom’s hand with one flat cooking spatula. Jerry carries nothing.
Camera: One continuous static shot. No cuts or camera movement.
0–3s: Jerry runs from right to left. Tom chases him and throws the spatula directly at Jerry. Show the spatula clearly leaving Tom’s paw and flying left in a straight, readable path.
3–5s: Jerry sees it coming and leaps upward. The spatula passes underneath him, misses, and flies straight into the third column of desktop icons. It hits the icons directly—it does not strike or bounce off the floor first.
5–7s: The six third-column icons—Gmail, Discord, Microsoft Teams, Microsoft Excel, Netflix, and Google Drive—come loose and scatter onto the floor. Their six grid positions are visibly empty. The other icons, menu bar, and centered Dock remain still.
7–11s: Tom stops, realizes what he has done, and rushes to the fallen icons. He picks up all six scattered icons from the floor and gathers them against his chest, holding them securely in both arms. Show the floor become clear as he collects them. Jerry watches nearby.
11–17s: Holding the remaining icons against his chest, Tom takes one icon at a time and places it into its correct empty position, from top to bottom: Gmail, Discord, Microsoft Teams, Microsoft Excel, Netflix, Google Drive. After each placement, that icon leaves his paw and stays in the grid; the number held against his chest decreases by one. Nothing snaps back by itself.
17–18s: Tom checks the restored column and gives Jerry an embarrassed glance. Jerry looks amused. End with all 18 icons in their original positions.
Critical action order: Tom throws the spatula → Jerry jumps clear → spatula directly hits the third column → six icons fall → Tom gathers all six against his chest → Tom replaces them one by one. Do not skip, reverse, or combine these actions. No floor ricochet, frying pan, cheese, popcorn, duplicated icons, changing logos, extra limbs, dialogue, subtitles, or camera movement.
Made this Pixar-style animated short with MiniMax H3 dev ✨
The price just dropped from 40% → 30%, making it even more affordable to create high-quality animated shorts.
Full tutorial shared below 👇
prompt:
Create an intense, fast-paced 2D cinematic motion-design sequence inspired by Attack on Titan. The anime title is for visual reference only: never show, write, hint at through typography, or reveal the title on screen. Do not use character names, logos, answer cards, explanatory text, or major plot spoilers.
Build the film as 15 distinct visual beats, escalating from ambiguous imagery to clues fans will recognize. Do not number or label the beats on screen.
1. A single feather drifts through a vast, oppressive sky; its shadow becomes a crack.
2. The crack races across weathered stone, revealing the immense curve of a fortified wall.
3. Dust lifts from empty rooftops as a distant, inhuman silhouette passes beyond the wall.
4. A tight mechanical close-up: cables snap taut, metal hooks fire, gas bursts into white vapor.
5. The vapor becomes sweeping speed lines as a cloaked silhouette rockets between narrow buildings.
6. A sharp flash of paired blades cuts through darkness; ink strokes trail behind the motion.
7. Falling masonry transforms into a flock of black birds crossing a blood-red sunset.
8. The birds’ wings match-cut into a dark cloak whipping violently in the wind.
9. A close-up of an eye reflects towering stone, moving clouds, and an enormous shadow.
10. Vertical rooftops and cables collapse into a dizzying aerial chase, drawn with rough anime linework and violent perspective.
11. A colossal hand emerges through steam as tiny human silhouettes scatter across a fractured street.
12. Steam expands into layered screen-print textures; silhouettes, blades, feathers, walls, and cables collide in rapid graphic flashes.
13. The flashes accelerate into aggressive whip pans and impact frames: hooks firing, boots landing, blades turning, stone breaking.
14. Earlier motifs remix into a dense 2D crescendo—red sky, black wings, white steam, monumental walls, and figures suspended in midair.
15. Everything cuts to near silence: one small silhouette stands before an immense wall while a vast shadow begins to rise behind it. Cut to black before the source of the shadow is shown.
Make every beat compositionally different. Move between minimal symbolic frames, extreme detail shots, painted environments, bold silhouettes, abstract transformations, and explosive anime-style action. Let each transition transform a shape or motion from the previous clue into the next.
Visual style: premium hand-drawn 2D anime animation fused with experimental editorial motion design. Use expressive ink linework, cel-shaded figures, painted shadows, rough brush strokes, weathered paper grain, halftone fragments, screen-print textures, torn graphic cutouts, dramatic negative space, sharp geometric framing, speed smears, and occasional one-frame impact images. Avoid clean flat-vector explainer aesthetics and avoid 3D CGI.
Color palette: charcoal black, weathered stone gray, parchment beige, muted military green, bone white, and restrained bursts of deep crimson. Keep the atmosphere urgent, monumental, gritty, and tragic without depicting explicit gore.
Use rapid composition changes, whip transitions, smash cuts, match cuts, graphic wipes, layered 2D parallax, sudden close-ups, aggressive scale changes, and rhythmic impact frames. The final beats should accelerate dramatically. End abruptly on an unresolved image. No title reveal, no confirmation, no guessing-game text.
Walls, grappling cables, flashing blades—15 shots that keep building momentum.
This Attack on Titan–inspired 2D motion video was generated with MiniMax H3 dev. It’s now 70% off, down from 60% off—a great time to try it.
@atlas_cloud_ai
prompt:
After school at sunset, a schoolgirl walks alone through a quiet old Japanese neighborhood, past tiled rooftops, power lines, bicycles, glowing vending machines, and laundry stirring in the evening breeze. A bright red omamori charm hangs from her school bag.
A black cat appears on a low wall and watches her for a moment. As she passes, it springs down, snatches the omamori firmly in its mouth, and runs. She turns and gives chase immediately.
The cat darts through narrow alleys, weaving between bicycles and flowerpots. The girl nearly corners it beside a vending machine, but it jumps onto the machine, crosses a row of outdoor AC units, and reaches the rooftops. She vaults a low wall, climbs a fire ladder, and follows.
They race across tiled roofs, jumping narrow gaps and dodging antennas, chimneys, water tanks, and laundry racks. The girl repeatedly draws close, but the cat slips just beyond her reach, the red omamori securely clenched in its mouth throughout.
They burst through large white sheets billowing across the rooftops, vanish briefly behind the fabric, then emerge at full speed. Ahead, the cat leaps across a wider rooftop gap against the orange-gold sky. The girl jumps after it and reaches for the omamori, missing by inches.
They land on the next roof and keep running deeper into the neighborhood until the final frame. The cat never drops or releases the omamori. No capture, no slowdown, no final pose.
Style: high-quality Japanese 2D animation with clean hand-drawn linework, natural anime proportions, detailed residential backgrounds, cel shading, warm golden-hour lighting, and soft blue-gray shadows. Fluid continuous running, climbing, and jumping with strong momentum. Hair, skirt, bag straps, laundry, and the omamori react naturally to wind and motion. Dynamic tracking camera, no manga panels, no text effects, no paper texture, no 3D CG.
At sunset after school, a black cat snatches the red omamori from a girl’s school bag. The chase takes them through quiet alleys and across the rooftops. 🐈⬛🌇
Generated directly with MiniMax H3 dev, The price drops from 40% to 30% of the original.
@atlas_cloud_ai
prompt:
Create a 30-second, 16:9, 720p live-action vampire drama trailer. It must feel like footage shot with real actors on a real film set, with the restraint and texture of a serious feature film.
Characters: The bride is a Western European woman in her late 20s. Brown hair loosely pinned under an ivory veil, faint freckles, natural skin texture, an intelligent, guarded gaze. She wears a simple ivory satin wedding dress and a plain silver ring. The groom is a Western European man in his early 30s. Dark, slightly untidy hair, a distinctive face with natural asymmetry, a black wool tuxedo with a muted burgundy lapel. He is quietly charismatic, never smug or seductive.
Setting: An old stone chapel at night. A few real candles, dark red roses, cool light from a high window. The space feels lived-in and imperfect. Handheld and restrained dolly shots, believable depth of field, natural shadows, subtle film grain. Keep faces, costumes and chapel consistent across cuts.
0–7s: During the wedding vows, the groom slips the ring onto the bride’s finger. She notices that his face is missing from a small antique mirror beside the altar. He notices her reaction. Their smiles fade. A church bell rings.
7–15s: The groom moves close to her neck. A brief glimpse of one fang. The bride appears trapped; her breathing quickens. Hold on their faces, not fantasy effects. Let the audience believe she is his prey.
15–23s: She suddenly stops trembling. With a precise movement, she locks his wrist and reveals a narrow silver blade hidden in her ring. She whispers, “I knew before you proposed.” He looks genuinely shocked. A sharp cut to silence.
23–30s: She presses the blade to his chest and says, “I came for the one who made you.” The groom’s fear turns into a small, involuntary smile. Behind her, an unseen figure exhales in the darkness. The bride turns. Cut to black before the figure is revealed. End on one heavy heartbeat.
Tone: tense, intimate, dramatic and grounded. The reversal must be clear through acting, timing and editing. Sparse music, natural chapel acoustics, clean English dialogue.
Avoid glossy beauty lighting, airbrushed skin, oily or wet-looking faces, overstyled hair, seductive posing, theatrical fangs, exaggerated red eyes, glowing magic, slow-motion glamour shots, CGI-looking architecture, symmetrical poster compositions, excessive blood, subtitles, text, logos and watermarks.
She thought she was marrying a vampire.
He thought he was marrying his next victim.
Only one of them knew what this wedding was really for.
A 30-second vampire drama trailer made with Seedance 2.5. I wanted it to feel like a live-action film: natural performances, restrained lighting, and a twist that lands in the final seconds.
Would you watch the full story? 🩸
Would you really make it through this rainforest without looking up at the trees?
This first-person rainforest scene was generated with Seedance 2.5, with the feel of handheld footage shot at night.
A head pat for Anya, then one for Bond when he runs over 🥹🐾
The character images were made with GPT Image 2.5’s free credits, and the video with GPT Image 2.5 + MiniMax H3 dev. MiniMax H3 dev went from 40% to 30% of the regular price, making this the cheapest combo right now!
@atlas_cloud_ai
prompt:
Use the provided first-frame image as the visual reference. Create a 10-second video in a 1:1 square format. Keep a fixed, first-person camera angle looking slightly down into the warm, sunlit bedroom. Preserve Anya’s and Bond’s appearance, the room layout, and the lighting throughout.
0–4 seconds: A hand gently enters from the bottom of the frame and pats Anya’s head. She blinks, smiles, and leans slightly into the touch. Her hair and ornaments move subtly.
4–6 seconds: The hand lifts away. Bond happily runs in from behind Anya and stops beside her, looking up at the camera and wagging his tail. Anya glances at Bond, then back at the camera.
6–10 seconds: The same hand reaches over and gently pets Bond’s head. Bond closes his eyes contentedly and keeps wagging his tail while Anya watches with a happy expression. End with both of them looking up at the camera.
Keep the motion soft and continuous, with natural hand-to-character contact and consistent character designs. No camera cuts, dialogue, subtitles, or additional characters.