- Category
- Video
River-Stone Figures Play Keepy-Uppy by a Waterfall
Create a photorealistic, single-take vertical video of a woman sharing a playful water-droplet game with two self-assembling river-stone figures beside a waterfall.
By Threads — Published on
Example results
Prompt template
- Format
- Text
- Inputs
- character_reference (image) · Required
- Outputs
- generated_video (video)
VIDEO PROMPT
Create one continuous, real-time 15-second video in 9:16 vertical format. Do not cut.
SCENE
Beside a waterfall pool, an adult woman eats lunch on a flat rock. Two small humanoid figures made of smooth river stones assemble themselves at her feet and play keepy-uppy with one water droplet. She joins the game. The figures eventually collide and collapse into ordinary stones.
CHARACTER REFERENCE
Use the supplied character reference image as @beauty; her face must match it throughout. She is an adult woman in her mid-20s, slim, with large dark brown eyes, freckles across her nose and cheeks, full lips, and dark brown hair in a braided low bun with loose face-framing strands and a thin red ribbon. Her voice is bright, warm, slightly hushed, and mid-register.
LOCATION AND COMPOSITION
Foreground: wet fern fronds and a mossy rock edge at the left and bottom of frame, heavily defocused into soft green bokeh. Midground: a broad, dark, wet, flat granite rock half-submerged in a clear shallow pool. @beauty sits cross-legged in the upper two-thirds of the frame; the two stone figures stand on the rock in the lower third, about one meter in front of her. Background: a waterfall about ten human-heights tall pours into the pool, churning white at its base, with 30% mist density and jungle canopy behind it. Soft overcast light filters through the leaves.
The camera is low at water level on the shadow side, looking slightly up at her and level with the stone figures. The first frame is already active: @beauty sits cross-legged with an open bento box balanced on one knee, chopsticks halfway to her mouth, mid-chew, eyes drifting toward the waterfall. Her bare feet rest at the water's edge. The glistening rock in front of her feet is empty. Her head and shoulders occupy the upper third, bento and hands the middle, and empty foreground rock the lower third.
CAMERA AND OPTICS
Handheld-smartphone feel, camera held 20 cm above the water with a natural 1 cm breathing sway. Slowly push in at 2 km/h for 30 cm over the take, easing to a stop during the final three seconds. Hold a 47-degree observational field of view. Shallow depth of field; rack focus down to the figures at 2.5 seconds and back to her face at 13 seconds. Focus rides between her face and the figures. Ferns create heavy foreground bokeh across 25% of the frame. Resolve macro-level detail in the stones and droplet. Use wide tonal latitude and soft highlight roll-off on the white water.
ACTION AND TIMING
0.0–2.5 seconds: @beauty chews slowly, looks toward the waterfall, and lowers her chopsticks toward the bento. On the wet rock in the lower third, loose river stones begin to shift. Seven smooth gray stones lift and assemble into a standing figure about as tall as the span of her open hand: a large oval head, three stacked rounded torso stones, and smaller pebbles forming jointed arms and legs. Rack focus down to it.
2.5–4.5 seconds: A second, slightly broader and one-stone-shorter figure assembles beside the first from mixed dark and light stones. @beauty freezes with chopsticks suspended, lips parting and eyes widening; she leans forward 15 cm as the bento tilts on her knee. The first figure crouches. One perfectly round water droplet, as wide as its stone head, rolls to rest against its foot.
4.5–7.0 seconds: The first figure taps the droplet upward with a stone-foot tip, juggles it in three small bounces between its feet, catches it on a raised stone knee, then flicks it to the second figure. The second traps it, bounces it twice, and heads it back. They pass it between them in an unhurried rhythm, each reacting a beat late. @beauty's shock turns into a widening open-mouthed grin; without looking away, she sets the bento on the rock beside her.
7.0–10.5 seconds: She slowly extends one index finger into the lower frame, palm up. The second figure knocks the droplet toward her. She flicks it back with her fingertip, laughing silently, shoulders shaking. They pass it in a triangle—finger, stone foot, stone knee, finger. The droplet arcs 20 cm each time, wobbling and re-forming on every touch. She grows bolder and taps it with the side of her thumb.
10.5–13.0 seconds: She flicks the droplet slightly too far. Both figures lunge for it, swing their stone arms for balance, and clip shoulders. They overbalance and come apart mid-stride; their stones tumble onto the wet rock and clatter into two loose piles. The droplet lands between the piles and bursts into the shallow water, sending out ripples. Her hand stops mid-air.
13.0–15.0 seconds: Stillness. She looks at the two piles of ordinary stones, then at the settling ripples, and her expression softens. Rack focus back to her. She looks into the lens, presses her lips together, and breaks into a small delighted smile, cheeks lifting and one shoulder rising. At 13.8 seconds, she says quietly, half-laughing, straight into the lens: “That is cool!” End on her held smile with ripples still moving in the lower frame.
PERFORMANCE AND PHYSICS
Render pore-level skin realism: mist beading on her cheekbones and forearms, a damp strand stuck to her temple, capillary flush high on her cheeks, and living eyes with soft overcast catchlights. Her reaction unfolds in order: chopsticks stop, then mouth opens, then eyes widen. Her final smile arrives slowly and reaches her eyes; her breathing remains visible in her shoulders throughout.
Give the stone figures slightly imperfect stop-motion weight: shifting stances, arms swinging for counterbalance, and a half-beat delay on each reaction. Include tiny grit sounds as stones seat against one another. Each stone retains real mass and remains in contact with the stone below it. The droplet maintains surface tension: it deforms and wobbles on contact, re-rounds in the air, carries a bright specular highlight, and stays whole until its final landing. Tiny ripples spread beneath every footstep. All three reflections—hers and those of both figures—move correctly in the still water.
LIGHTING AND WARDROBE
Fixed, soft, diffused overcast daylight from directly above and slightly behind the waterfall at 6000K. Bounce fill from the white churn lifts the shadow side of her face and the stone figures. Wet stone has a hard specular sheen along its top edges; the droplet carries one bright moving highlight on every bounce. Maintain a 6000K white balance.
She wears a loose cream linen shirt with sleeves rolled to the elbow, dark rolled-up hiking trousers, and no shoes. A thin canvas daypack sits behind her on the rock. The fabric hem is damp from pool spray.
AUDIO AND STYLE
Include steady waterfall roar, close water lapping, jungle birds, small dry stone-on-stone clicks with every figure step and touch, and one soft plink when the droplet lands. No other spoken line. Use an ultra-realistic live-action environment with photorealistic stone stop-motion characters and a physically simulated droplet. Tactile stone texture, natural muted greens and grays, cinematic macro depth of field, tangible handcrafted movement, and fine grain.
CONTINUITY LOCKS
Show exactly one woman and exactly two stone figures. Build the figures from visibly different stones so they remain distinguishable. Show exactly one droplet, whole and round from 4.5 seconds until it lands at 12.8 seconds. Keep each figure’s stone count and arrangement consistent from assembly through collapse. From 2.5 seconds onward, keep the woman and both figures in frame: she stays in the upper two-thirds and they stay on the rock in the lower third. Keep the waterfall, pool, ferns, and lighting identical throughout. Preserve the reaction order: shock, delight, then participation. Speak the line once, at the end, into the lens. No subtitles, watermark, or logo.
Comments — 0