Wan 3.0 Live-Action Comedy Short: 9-Shot Direction and UI Physicalization Prompting Guide

AI video creator John (@johnAGI168) has published an advanced prompt engineering framework using the Wan 3.0 video foundation model to orchestrate a 30-second, 16:9 horizontal live-action comedy chase short. The technique's core breakthrough lies in physicalizing familiar image-editing GUI elements—such as the mouse cursor, marquee selection box, paint bucket, eraser, and transparent checkerboard grid—into tangible real-world props embedded directly within a sunlit urban street, complete with perspective, occlusion, realistic cast shadows, and frame-accurate audio cues.

Image source: X @johnAGI168

While typical AI video prompts are restricted to single-subject scenes or ambient landscape drifts, this workflow subdivides the entire 30-second narrative into nine structured shots. By combining specific focal lengths (24mm, 35mm, 50mm) with continuous physical momentum and strict negative constraints, the prompt achieves coherent pacing and slapstick comedy timing directly within a single structured prompt.

GUI Physicalization and Overall Cinematic Direction

The foundation of the prompt is treating 2D software interface controls as physical three-dimensional objects interacting with actors and environment geometry.

• Format & Aspect Ratio: 30 seconds total duration, 16:9 horizontal widescreen format.
• Visual Style: Cinematic live-action daytime urban comedy. Featuring real human actors, authentic commercial street locations, and the dynamic energy of handheld tracking camera movement.
• Physical Attributes of UI Elements: The mouse cursor, dotted marquee selection box, transform bounding box, white square corner handles, and gray-and-white transparent checkerboard grids exist as physical items on the street. They obey perspective rules, exhibit realistic foreground/background occlusion, and cast soft gray shadows onto the pavement.
• Physical Momentum Rules: Every visual manipulation triggers instantly and cleanly. Character and object motion preserves continuous physical momentum across cuts, strictly prohibiting awkward idle pauses or synthetic, plasticky CGI artifacts.

Subject Binding, Spatial Axis, and Frame-Accurate Audio Timeline

To prevent model drift across multi-shot transitions, the prompt establishes rigorous character references, prop behaviors, directional axes, and frame-accurate audio synchronization.

1. Subject and Prop Binding

• Main Character (Iaro): Bound to a single reference image (Image 1). Facial features, haircut, apparel, and accessories remain completely consistent throughout the entire film.
• Prop [Cursor]: A giant physical computer cursor with a subtle black outline and a matte white finish. Standing approximately 1.2 meters tall, it floats between 1.0 and 1.8 meters above the ground, casting a faint gray shadow. Given a stubborn, mischievous personality, the cursor shifts instantly ("snap" transformation) among three functional states:
  • Arrow Form: The classic white arrow pointing toward the upper-left.
  • Paint Bucket Form: A tilted white bucket icon with the spout pointing downward.
  • Eraser Form: A solid rectangular white eraser block approximately 1 meter in length.
  • On every "click," the cursor propels forward roughly 10 centimeters with a crisp mechanical clicking sound.
• Supporting Characters and Animals: The dog walker is a young woman wearing a light blue hoodie and holding a red leash. Xiaobai is a small curly-haired white dog with a standard shoulder height of 25 centimeters, scaling proportionally up to 1.5 meters into a cheerful giant dog that playfully pursues moving targets.

2. Spatial Environment and Chase Axis

• Lighting & Setting: A modern commercial street on a clear day. Natural sunlight falls from the upper-right at a 45-degree angle (color temperature around 5600K), casting soft yet distinct shadows. The sidewalk features wooden benches, a parked white box delivery van, a red brick wall, and a dead-end narrow alley lined with brick walls.
• Consistent Chase Axis: Chase orientation remains unified throughout the film: Iaro always flees from screen-right to screen-left, and the cursor always pursues from screen-right. Storefront signs, vehicle bodies, and wall surfaces are strictly devoid of letters, numbers, or logos.

3. Frame-Accurate Audio Timeline (0.0s–30.0s)

The short is structured as dialogue-free visual slapstick. A low-level funk track provides rhythm, synchronized with exact audio timestamps:

• 0.0s–: Urban background ambience blends into a lighthearted original funk soundtrack entering at low gain.
• 1.2s–2.4s: Four rapid "pop, pop, pop, pop" sounds as four ice cream scoops are duplicated.
• 4.0s / 5.6s: First mechanical "click" at 4.0s; second "click" at 5.6s as the bench duplicates; followed by a wet "splat" at 6.2s as ice cream scoops hit the pavement.
• 9.4s: A "click" followed by an elastic stretching sound ("boing-"); answered at 10.2s by a deep, resonant bark ("woof!") from the enlarged dog.
• 13.2s: A rushing cascade of liquid paint ("whoosh-").
• 16.2s: High-friction eraser rubbing against brick ("scritch-scratch").
• 19.0s: Low resonant hum as the cursor expands ("whooo-"), followed by playful chewing and squeaking sounds at 20.0s.
• 23.5s: A wet shake as the cursor shakes off dog saliva ("shudder-flutter").
• 25.5s: Music abruptly cuts into dead silence.
• 25.8s / 26.0s: Sharp "click" at 25.8s; sudden "pop" at 26.0s as Iaro's hair transforms into a red afro; topped by a bouncy spring sound ("boing") at 26.8s.
• 28.0s–30.0s: Ambient silence punctuated by the cursor's playful electronic chirp ("beep-beep") and the giant dog's tail thumping the pavement ("thump, thump, thump").

9-Shot Sequence Breakdown and Structured Prompt Reference

The 30-second narrative unfolds across nine tightly synchronized shots, governed by strict negative boundaries:

Shot	Timecode	Lens & Framing	Staging and Action
SHOT 1	0.0s–3.5s	35mm eye-level medium shot	[Ice Cream Duplication] As the protagonist sits on a wooden bench about to lick an ice cream cone, the floating cursor appears above it, drawing a dotted marquee box to duplicate four extra scoops.
SHOT 2	3.5s–7.5s	35mm eye-level	[Targeting Iaro] The marquee box targets the wooden bench and Iaro. A duplicated bench drops to the ground, jarring the ice cream scoops onto the pavement. Iaro slides sideways and breaks into a sprint.
SHOT 3	7.5s–12.0s	24mm tracking follow shot	[Xiaobai Becomes Giant] Iaro leaps over the small white dog. The transform bounding box snaps onto the dog, scaling it up proportionally to a 1.5-meter shoulder height.
SHOT 4	12.0s–15.0s	35mm eye-level	[Paint Bucket: Van Turned Purple] The cursor transforms into a paint bucket and clicks the parked white delivery van, spreading vibrant bright purple across the vehicle body like ripples in 0.6 seconds without liquid splashing. Iaro pushes off the vehicle to dodge.
SHOT 5	15.0s–18.0s	35mm medium shot	[Eraser: Transparent Grid on Wall] Transforming into an eraser block, the cursor sweeps across a red brick wall, erasing a 1-meter square into a floating gray-and-white checkerboard transparency grid.
SHOT 6	18.0s–22.5s	24mm low-angle to 35mm medium	[Dead End: Giant Dog Rescue] The cursor expands to 3 meters wide, blocking the narrow alleyway exit. The giant dog leaps in, snatching the cursor in its jaws and shaking it like a chew toy.
SHOT 7	22.5s–25.5s	35mm eye-level	[A Brief Respite] Iaro catches his breath and gently scratches the giant dog's chin. The saliva-covered cursor floats back up and shakes off water droplets.
SHOT 8	25.5s–28.0s	35mm slow push-in close-up	[Red Afro Transformation] The cursor snaps a marquee box over Iaro's head and clicks. His hair pops instantaneously into a perfectly spherical, 60cm bright red afro.
SHOT 9	28.0s–30.0s	50mm close-up freeze	[Are You Serious?] Iaro stares blankly at the cursor in disbelief while the giant dog fixates on the bouncy red hair like a tennis ball, crouching to pounce as the frame freezes.

To enforce production quality, negative constraints ban text, watermarks, subtitles, spoken dialogue, idle pauses, face or costume drift (outside Shot 8), plastic skin shaders, gore, cursor clipping through humans, and cartoon animation aesthetics. Below is the complete structured prompt ready for direct input into Wan 3.0 generation systems.

Duration: 30 seconds
Aspect ratio: 16:9 widescreen
Overall style: Daytime urban comedy with cinematic live-action texture. Real human actors, real commercial street, dynamic handheld follow-camera energy. GUI elements and tools from image-editing software appear as physical tangible objects on the real street, including mouse cursor, dotted marquee box, transform box, white square corner handles, and gray-and-white checkerboard transparency grids. They exhibit correct perspective, occlusion, and cast soft gray shadows onto the ground. All edits trigger instantly, cleanly, and playfully. Continuous physical momentum across all actions; strictly no idle pauses and no plastic CG artifacts.

SUBJECT / CHARACTER & PROP BINDING:
Image 1 designated as [Iaro]: facial features, haircut, apparel, and accessories strictly follow the reference image and remain consistent throughout the entire short.
Prop [Cursor]: A giant computer cursor as a physical matte white object with a subtle black outline. Standard height ~1.2m, floating 1.0 to 1.8m above ground with soft gray ground shadows. Three instant snap-transformation forms:
- Arrow form: Classic top-left pointing white arrow.
- Paint bucket form: Tilted white bucket icon with spout facing downward.
- Eraser form: Solid rectangular white eraser block ~1m long.
Every "click" snaps the cursor forward ~10cm accompanied by a crisp mechanical click sound. Stubborn and mischievous personality.

BYSTANDERS & ANIMALS:
[Dog Walker]: Young female wearing a light blue hoodie, holding a red leash.
[Xiaobai]: Small curly-haired white dog, normal shoulder height ~25cm, proportionally scaled up to 1.5m shoulder height when transformed. Cheerful and excited personality, eagerly chasing moving objects.

ENVIRONMENT / SPATIAL AXIS:
Modern commercial street on a clear sunny day. Natural sunlight from upper right at 45 degrees, color temperature ~5600K, soft distinct shadows. Wooden benches on the sidewalk, parked white box van, red brick wall, and a dead-end narrow alley between brick walls. Unified chase axis: Iaro always flees from screen-right to screen-left; cursor always pursues from screen-right. Strictly no text, numbers, or logos on signs, vehicles, or walls.

AUDIO / SOUND TIMELINE (No dialogue, no human vocals):
① 0.0s-: Urban street ambience, original upbeat funk soundtrack (bass and snare, no vocals) entering at low volume.
② 1.2s–2.4s: Four rapid "pop, pop, pop, pop" sounds as extra ice cream scoops duplicate.
③ 4.0s: "click"; 5.6s: second "click", bench duplicates with a "thud" landing; 6.2s: four ice cream scoops hit pavement with a "splat".
④ 9.4s: "click" followed by elastic stretch "boing-"; 10.2s: giant dog deep joyful bark "woof!".
⑤ 13.2s: Paint pouring cascade "whoosh-"; 16.2s: eraser rubbing brick "scritch-scratch".
⑥ 19.0s: Cursor expands with rising low frequency "whooo-"; 20.0s: dog barks and playfully chews cursor with squeaky toy sounds.
⑦ 23.5s: Saliva water-droplet flick sound "flutter-shudder".
⑧ 25.5s: Music cuts abruptly into dead silence; 25.8s: "click"; 26.0s: afro explodes open "bang!"; 26.8s: bouncy spring vibration "boing".
⑨ 28.0s–30.0s: Ambient silence, cursor cheeky electronic chirps "beep-beep", giant dog tail thumps pavement "thump, thump, thump".

SHOTS BREAKDOWN (1 to 9):
SHOT 1 (0.0s–3.5s): [Ice Cream Duplication] 35mm eye-level medium shot. Cursor marquee box duplicates 4 extra ice cream scoops.
SHOT 2 (3.5s–7.5s): [Targeting Iaro] 35mm eye-level. Marquee box locks onto bench and Iaro. Bench duplicates and lands, jarring scoops onto ground. Iaro slides sideways and sprints left.
SHOT 3 (7.5s–12.0s): [Xiaobai Becomes Giant] 24mm tracking follow. Iaro leaps over small dog. Transform box snaps onto dog, scaling it up proportionally to 1.5m shoulder height.
SHOT 4 (12.0s–15.0s): [Paint Bucket: Van Turned Purple] 35mm eye-level. Cursor turns into paint bucket, clicks parked delivery van; bright purple ripples across vehicle body in 0.6s without splashing. Iaro pushes off to dodge.
SHOT 5 (15.0s–18.0s): [Eraser: Transparent Grid on Wall] 35mm medium shot. Cursor turns into eraser, wipes across red brick wall, erasing a 1m square into a floating gray-and-white checkerboard transparency grid.
SHOT 6 (18.0s–22.5s): [Dead End: Giant Dog Rescue] 24mm low-angle to 35mm medium. Cursor expands to 3m blocking narrow alleyway. Giant dog leaps in, biting and playing with cursor like a toy.
SHOT 7 (22.5s–25.5s): [Brief Respite] 35mm eye-level. Iaro scratches dog's chin. Saliva-covered cursor floats up and shakes off water droplets.
SHOT 8 (25.5s–28.0s): [Red Afro Transformation] 35mm slow push-in close-up. Cursor boxes Iaro's head and clicks, instantly transforming hair into a 60cm spherical bright red afro.
SHOT 9 (28.0s–30.0s): [Disbelief / Freeze Frame] 50mm close-up. Iaro stares blankly at cursor. Dog fixates on bouncy red afro like a ball and crouches to pounce. Frame freezes.

NEGATIVE CONSTRAINTS:
No text, letters, numerals, logos, watermarks, subtitles. No spoken dialogue or vocalizations. No idle initial pauses. No character identity drift (except Shot 8 hair). No fake wig look, plastic skin, or wax-figure CG face. No blood, gore, horror, or aggressive dog behavior. No cursor clipping through human bodies; ground shadows mandatory. No chase axis reversal; no unmotivated floating objects; no cartoon/2D animation styling.

Sources

• John official X account (@johnAGI168): Wan 3.0 Comedy Chase Short Film Prompt Guide & Breakdown
• Video demonstration and frame-accurate audio timestamp reference: Wan 3.0 Live-Action Video Post