Subject
Name the main thing in the shot and one or two stable traits so it stays consistent across frames.
"a fox with a white-tipped tail"Muse Video Prompt Guide
Write prompts that give you controllable, cinematic clips. This guide covers the prompt formula, eight ready-to-use examples, the difference between text-to-video and image-to-video, and fixes for the most common problems.
The prompt formula
A strong clip prompt covers eight building blocks. You do not need all of them every time — but the more concrete each one is, the less the model has to guess.
Scene
Say where the subject is, plus one environmental detail that grounds the shot.
"in a snow-covered pine forest at dusk"Action
Give one primary action per clip — a single clear verb beats three vague ones.
"running toward the camera"Camera movement
Describe how the camera moves; this is the single biggest lever against chaotic motion.
"slow dolly-in, eye level"Lighting
Set the light direction and mood so the shot reads as intentional, not random.
"cold moonlight with warm lantern glow"Style
Anchor the look: cinematic, product ad, anime, stop-motion, documentary.
"cinematic, shallow depth of field"Pacing
Hint at tempo — calm and slow vs fast cuts and energetic motion.
"slow, deliberate motion"Format & length
Pick aspect ratio (16:9, 9:16, 4:3, 3:4), duration (5/10/15s) and resolution (480p/720p/1080p) in the generator — not inside the prompt text.
"9:16 · 10s · 720p for a social ad"8 copy-ready prompt examples
Each card shows a complete prompt plus the recommended settings. Try the cheap drafts first (480p · 5s costs 160 credits), then re-run your favourite at higher quality.
Cinematic city at night
A cinematic video of a night train gliding through rain-soaked city streets, reflections on wet asphalt, slow dolly-in camera, moody neon lighting, grounded realistic motion
Try this promptVertical product ad
A premium skincare bottle rotating slowly on black glass, blue light trails wrapping around it, smooth camera orbit, glossy launch-ad lighting, macro product detail
Try this promptCharacter close-up
Close-up of an astronaut slowly removing her helmet inside a glowing botanical lab, consistent face and suit, gentle handheld camera drift, shallow depth of field, quiet determined mood
Try this promptSocial vertical short
Paper cutout planets unfolding into a miniature solar system on a wooden desk, playful stop-motion energy, crisp shadows, one continuous move, bright flat lighting
Try this promptImage-to-video: bring a photo to life
Animate the uploaded photo: the subject stays identical, camera pushes in slowly, hair and clothes move in a light breeze, natural blinking, no scene change
Try this promptAerial nature flyover
Aerial drone shot flying low over a turquoise coastline, waves breaking on volcanic rock, golden-hour light, smooth forward glide, epic but realistic scale
Try this promptFood close-up
Macro shot of steam rising from a bowl of ramen, chopsticks lifting noodles, soft window light, shallow focus on the broth, warm appetizing tones
Try this promptAnime-style scene
Anime-style shot of a girl waiting on a train platform at sunrise, her scarf moving in the wind, cherry petals drifting past, gentle pan camera, soft pastel palette
Try this promptText-to-video vs image-to-video prompts
Text-to-video: the prompt carries everything — subject, scene, action, camera, light and style. Be concrete about all of them, because nothing is pinned down by an image.
Image-to-video: the reference image already fixes the subject and scene, so spend your words on what changes over time — the motion, the camera move, the atmosphere. Do not re-describe what is already visible in the image; it wastes tokens and can fight the reference.
Common problems and how to fix them
- The camera moves chaotically
- Name one camera move explicitly ("slow dolly-in", "static tripod shot", "gentle pan left"). If you write nothing, the model invents one — usually an unstable one.
- The subject changes between frames
- Anchor it with one or two stable traits, or switch to image-to-video with a reference image so identity is pinned by the picture.
- Too much happens at once
- One clip = one primary action. Split complex ideas into separate clips instead of cramming three actions into one prompt.
- The result looks generic
- Add one concrete detail per block — a named place, a specific light source, a lens feel — instead of adjectives like "beautiful" or "epic" alone.
- Vertical video looks cropped
- Choose 9:16 in the generator settings rather than describing vertical framing in the prompt; aspect ratio is a parameter, not prompt text.
Settings and credit cost
Generation is priced per clip by resolution and duration (1 USD = 100 credits). The exact cost is always shown in the generator before you run it.
| Resolution | 5s | 10s | 15s |
|---|---|---|---|
| 480p | 160 | 320 | 480 |
| 720p | 240 | 480 | 720 |
| 1080p | 360 | 720 | 1080 |
Prompt FAQ
- How long can a prompt be?
- Up to 2,000 characters. Shorter, denser prompts usually beat long ones that repeat themselves.
- Can I write prompts in Chinese or other languages?
- Yes — prompts are not limited to English. Write naturally in your language; the same formula still applies.
- Should I mention aspect ratio or duration in the prompt?
- No — the generator settings take precedence. Choose aspect ratio, duration and resolution in the controls; there is no need to write them into the prompt text.
- How do I reuse a prompt that worked?
- Open your generation history, find the clip you liked, and reuse or tweak its prompt from there — small edits beat rewriting from scratch.
- Why did my prompt get rejected?
- Some subjects may be blocked by the content-safety filter — see the Content Safety Policy page for details. Adjust the related wording and try again.