Muse Video Prompts: Complete Framework, Audio & Examples

A production-ready Muse Video prompt framework

A strong AI video prompt is a compact production brief. It explains not only what a frame looks like, but how the subject moves, how the camera responds, what must remain consistent, what the viewer hears, and how the shot ends. This matters for Muse Video because Meta specifically emphasizes prompt adherence, temporal consistency, visual fidelity, and native audio.

Use this structure:

Subject + action + environment + camera + continuity + lighting + pacing + audio + final beat

Do not add adjectives endlessly. Give every clause a production purpose.

The nine prompt components

1. Subject

Identify one primary subject precisely: appearance, material, clothing, product shape, or other details that must remain recognizable. Too many equal-priority subjects increase ambiguity.

2. Action

Describe an action that can unfold within the requested duration. “A runner accelerates, clears one hurdle, and slows at the finish” is more filmable than a list of unrelated events.

3. Environment

Define the location, time, weather, background activity, and spatial relationship between subject and surroundings. Environment should support the action rather than compete with it.

4. Camera

Specify shot size, angle, movement, speed, and lens character. Use one readable camera path: slow dolly-in, lateral tracking shot, controlled orbit, handheld follow, or locked macro shot.

5. Continuity

State what cannot change: face, logo, packaging, clothing, hand count, room geometry, vehicle shape, or direction of travel. Continuity instructions make evaluation objective.

6. Lighting and visual treatment

Describe the source and behavior of light rather than only saying “cinematic”: soft window light, warm backlight, cool overhead practicals, reflected neon, shallow depth of field, restrained film grain.

7. Pacing

Divide the shot into an opening, development, and final beat. Timing words such as “after two seconds” or “on the final beat” help coordinate motion and sound.

8. Native audio direction

Name the sound source, character, distance, and timing. Keep dialogue short. Separate ambience, effects, dialogue, and music so they do not fight for attention.

9. Final beat

Give the clip a deliberate ending: the product reaches its mark, the character looks into camera, the title sound resolves, or the camera settles into a clean final frame.

Complete Muse Video prompt examples

Cinematic wildlife

A solitary white owl glides between tall pine trees at blue hour. Low tracking camera follows beneath the wings, then slowly pushes closer without overtaking the bird. Preserve the owl's white facial markings and wing shape throughout the shot. Cool moonlight, soft rim light, realistic feather motion, restrained depth of field. Quiet opening, gradual acceleration, then a stable final glide above a misty clearing. Native audio: close wing beats, distant wind through pine needles, subtle forest ambience; no music or narration.

Product advertisement

A matte-black wireless speaker stands on a wet stone platform in a dark studio. The camera performs one slow 120-degree orbit as narrow warm lights travel across the grille. Keep the logo, button layout, dimensions, and surface material unchanged. Fine water droplets vibrate subtly with the bass. Premium commercial lighting, crisp reflections, deep black background. Native audio: one low bass pulse at the opening, soft electrical texture during the orbit, clean sonic logo exactly as the camera stops on the final front view.

Dialogue scene

Medium close-up of a tired radio host in a quiet midnight studio. He glances at the empty guest chair, takes one breath, and says: “Some stories arrive after the signal fades.” Locked camera with a very slow push-in. Preserve facial identity, jacket, microphone position, and background practical lights. Warm key light against cool blue shadows. Native audio: intimate dry voice, low room tone, faint vinyl crackle; no music. End on two seconds of silence after the line.

Vertical social hook

Vertical 9:16 macro shot of a translucent running shoe landing on a reflective floor. On impact, colored layers separate briefly to reveal cushioning, support, and outsole, then reconnect into the original shoe. Fast but readable camera move, stable branding and exact product proportions. Bright laboratory lighting with crisp colored edges. Native audio: sharp landing impact, three short material clicks synchronized to the separated layers, upbeat final sting. Hold the complete product centered for the final second.

Image-to-video direction

Animate the supplied portrait without changing the person's face, hairstyle, clothing, jewelry, or background architecture. Add a gentle breeze to loose hair and fabric, a subtle shift of sunlight, and one slow camera push-in. Natural blinking only once. No talking, no exaggerated smile, no new objects. Native audio: light rooftop wind and distant city ambience. End on the same composition as the reference, slightly closer.

Native-audio checklist

• Identify what produces each important sound.
• Say whether dialogue is on-camera, off-camera, whispered, distant, or dry.
• Keep spoken lines short enough for the clip duration.
• Assign synchronization points to impacts, cuts, gestures, or the final reveal.
• Avoid loud dialogue, dense music, ambience, and many effects at the same time.
• Treat lip sync and fast-motion sound timing as test criteria; Meta says synchronization remains under improvement.

Prompt review before generation

Before spending credits, confirm that the prompt has one main subject, one readable action, one camera path, explicit continuity constraints, a clear ending, and audio that serves the scene. Save the exact prompt and model identifier with every output. Change one variable at a time when comparing results.

Use these prompts in the AI video generator, examine the official preview examples, or return to the complete Muse Video guide.

Source and scope

This framework is editorial guidance derived from standard video-production practice and the capabilities and limitations described in Meta's official Muse announcement. It is not an official Meta prompt specification.