FilmLune · AI generator and gallery
Curated prompts | Create AI images & videos with ease
Understand FilmLune
FilmLune combines an AI image generator with a gallery of image and video examples. Each detail shows the result, available prompt, declared model, and a method you can adapt.
Explore all promptsA readable method
Two ways in
Start with an image example for a still composition, or a video example to study motion, inputs, and duration. Both libraries lead to the same canonical case details.
Useful boundaries
Models and settings describe the visible revision: FilmLune does not silently replace a model during handoff. External work keeps its attribution and source link on the detail page.
Only when that revision’s rights permit it. The detail then preserves the prompt and its provenance; an unavailable action is never replaced with invented content.
No. Model versions, inputs, and settings can change the output. Use the method as a starting point and check the parameters before generating.
SpringWangSeedance 2.5PROMPT
Create one 9:16 vertical image with the realistic look of a spectator’s smartphone photograph, as the opening frame of a playful, physically impossible VFX stunt.
The setting is a large modern shopping mall with a tall, open central atrium extending several floors down. Multiple levels of shops and glass-fronted balconies surround the void. Long escalators run diagonally through the left side of the background. Above is a high glass ceiling with a rectangular metal grid. Far below are small retail kiosks with green rectangular roofs, bright walkways, and tiny shoppers.
On the right half of the image, a clearly adult woman in her mid-twenties stands on top of the rounded silver metal handrail of an upper-floor glass balustrade. She has dark hair in a high ponytail and wears a light cream short-sleeved cropped T-shirt with fine gray horizontal stripes, matching casual shorts, and thick white mid-calf slouch socks.
IMPORTANT POSITION: both of her sock-covered feet are visibly resting on the narrow TOP of the silver handrail, not on the walkway or a broad ledge. Her full body is visible from head to feet. She faces screen-left toward the open atrium, seen in a three-quarter side view. Her legs are nearly straight, her arms rest beside her body, and she looks down into the atrium. This is the quiet moment immediately before takeoff.
The photographer is on the same upper-floor walkway, a few steps behind and to her side. The handrail extends diagonally from the lower foreground toward the middle-right. The public walkway and a small group of watching shoppers remain visible along the far-right edge.
Through the glass and near the lower-left edge of the frame, reveal a small, partially cropped view of a large circular trampoline on the ground floor, several stories below. It has a black jumping surface and a thick bright orange padded rim. It sits in a clear area of the atrium, slightly outward from the balcony, with spectators gathered loosely around it.
Natural indoor mall lighting, believable architectural perspective, ordinary smartphone wide-angle framing, realistic skin and fabric. Emphasize the deep drop and the contrast between the nearby woman and the tiny ground-floor details.
One complete scene. No captions, watermark, logos, or video-player interface.Create a 10-second, 9:16 vertical image-to-video clip using the supplied image as the exact opening frame.
Style: a playful, physically impossible trampoline stunt presented as realistic handheld smartphone footage inside a shopping mall. One continuous take at normal speed.
Preserve the same adult woman, ponytail, striped outfit, white socks, silver handrail, glass balustrade, escalators, and mall architecture throughout.
CAMERA RULE:
The person filming stays on the original upper-floor walkway for the entire video. The phone tilts down to follow the descent, then tilts back up to follow the rebound. The camera never travels down into the atrium. Keep slight natural handheld movement without losing the performer.
[0.0–1.0 seconds]
Starting from the supplied frame, the woman briefly bends her knees and makes one small outward jump from the silver handrail into the open atrium. Both feet clearly leave the rail. Her ponytail lifts with the movement. The phone begins tilting downward after her.
[1.0–3.6 seconds]
She descends rapidly, feet first, through the clear space beside the escalators. Her body remains mostly upright, with her arms slightly spread. She becomes smaller as she moves farther from the upstairs camera.
The phone tilts into a steep downward view, revealing the full circular trampoline below: a black mat surrounded by a thick orange padded ring, with spectators standing around the outside. Keep her descending body clearly visible and aligned with the center of the mat.
[3.6–4.5 seconds]
Both sock-covered feet land in the CENTER of the black trampoline mat. The mat visibly stretches downward into a deep bowl beneath her weight, and her knees bend.
The stretched mat immediately recoils upward, launching her into one exaggerated, powerful rebound. Show the complete contact, compression, and release as a continuous movement, synchronized with a deep elastic THUMP-BOING.
[4.5–7.5 seconds]
The rebound sends her rapidly back up through the same open atrium toward the original balcony. During the ascent, she pulls her knees toward her chest and performs ONE complete tucked backward somersault.
The phone tilts upward to follow her. She grows larger as she approaches the camera’s floor, and the original escalators and upper balconies return into view.
After completing the rotation, she opens out of the tuck and extends her legs downward, returning to an upright orientation. Her upward path curves gently back toward her original position on the handrail.
[7.5–8.8 seconds]
She reaches the original upper-floor railing and lands with BOTH feet on the EXACT SAME SECTION of the silver handrail where she started.
Show visible foot-to-rail contact. Her knees bend into a brief crouch and her arms extend for balance. Her ponytail swings forward from the landing. The phone settles back toward its original framing.
[8.8–10.0 seconds]
She straightens fully and stands steadily on the handrail in almost the same pose as the opening frame, facing the atrium. She lowers her arms and gives a small, satisfied smile.
Hold this completed return long enough to make it unmistakable: she is back on the original rail, balanced and stationary. Nearby spectators react with excited screams and cheers.
AUDIO:
Quiet, reverberant shopping-mall ambience throughout. A brief air-and-clothing swish during the drop, one clear trampoline THUMP-BOING during compression and release, and spontaneous crowd gasps, screams, and cheers during the rebound and final landing. No music, narration, or spoken dialogue.
CONTINUITY:
One outward jump, one trampoline landing and rebound, one complete somersault, and one return landing. The trampoline stays fixed on the ground floor. The final landing is on the original handrail, not the walkway. Continuous forward action, not reversed footage. No cuts, slow motion, captions, watermark, or player interface.A realistic spectator-photo opening frame shows a young woman standing on a mall handrail above a multi-story atrium. The ten-second clip that follows sends her jumping off, rebounding from a ground-floor trampoline, performing one back somersault, and landing back on the exact same rail.
Video prompts and examplesThis image-plus-video pairing builds a playful, physically impossible VFX stunt that keeps the look of authentic handheld smartphone footage. The still frame establishes the setting, the vertical scale, and the performer's precise position; the video uses that frame as its opening shot for a complete out-and-back journey between an upper balcony and the atrium floor, with no cuts and no slow motion. The intended feel is a spontaneous bystander capture with the continuity of a true single take.
Start with the image: a 9:16 vertical frame shot from spectator height on an upper walkway of a large modern mall. The atrium drops several floors, ringed by glass-fronted balconies, with escalators cutting diagonally through the left background, a metal-grid glass ceiling overhead, and green-roofed kiosks and tiny shoppers far below. On the right half, an adult woman in her mid-twenties, dark high ponytail, cream striped cropped tee, matching shorts, and thick white slouch socks, stands with both sock-covered feet resting on the narrow top of the rounded silver handrail, three-quarter view, facing the void and looking down. The circular trampoline with a bright orange padded rim peeks into the lower-left edge, partially cropped, several stories down. Then the video: ten seconds, same vertical format, using the supplied image as the exact first frame. From 0 to 1 s she bends her knees and makes one small outward jump off the rail; from 1 to 3.6 s the phone tilts down to track her feet-first descent and reveals the trampoline below with spectators around it; from 3.6 to 4.5 s her feet hit the center of the black mat, which stretches into a deep bowl and recoils into one exaggerated rebound; from 4.5 to 7.5 s she rockets back up, pulls one complete tucked backward somersault as the phone tilts up, then opens out and returns upright; from 7.5 to 8.8 s she lands on the same section of the handrail, crouching briefly with arms out for balance; from 8.8 to 10 s she straightens, smiles, and stands still on the rail while onlookers cheer. The camera never travels into the atrium: it only tilts down and back up, with slight natural handheld sway. Audio stays minimal: quiet reverberant mall ambience, a brief swish during the drop, one elastic THUMP-BOING at compression and release, and spontaneous crowd gasps and cheers. No music, captions, or on-screen text.
The stunt depends on an exaggerated rebound, so the physics are deliberately unrealistic; the realism comes from the image texture, not the mechanics. Face, clothing, and hand consistency during the tucked rotation can drift between generations, and the final foot-to-rail contact often takes several attempts to land cleanly. The viewpoint stays fixed at the upper floor: any camera invention that descends into the atrium or drifts sideways breaks the filmer's perspective. Duration is fixed at ten seconds in vertical format, which leaves little room to slow or extend one phase of the motion without shortening the return landing.