78 Curated Seedance Prompts

Seedance 2.5 is strong for cinematic AI video: coherent motion, camera moves, and stylized sequences. A good seedance prompt is explicit about subject, action, and shot language.

Use this seedance prompt template and swap only the bracketed parts: [subject], [action], [environment], [shot/framing], [camera move], [lighting], [mood], [duration/pacing]

These seedance prompt examples prioritize clarity over adjective stacking:

• a lone cyclist, rides through light fog on an empty bridge at dawn, wide shot, slow tracking shot, soft blue hour lighting, calm cinematic mood, 5 seconds
• a chef, flips noodles in a wok, busy street food stall, medium shot, handheld slight shake, warm tungsten lighting, energetic mood, 4 seconds
• a black sports car, accelerates on a wet road, low-angle close-up, lateral tracking, strong rim light, high contrast, 3 seconds
• a dancer, performs a single spin, dark studio background, full-body shot, locked-off camera, softbox key light + rim light, elegant mood, 4 seconds
• a robot dog, walks toward camera, clean industrial hallway, centered shot, slow dolly-in, cool fluorescent lighting, subtle suspense, 5 seconds

A reliable seedance prompt uses a compact structure:

• subject identity (who/what)
• one primary action (avoid competing verbs)
• environment as a constraint (one location)
• camera (framing + movement)
• lighting + mood
• duration / pacing (if supported)

Most failures come from too many actions or unclear camera instructions. Fix your seedance prompt by:

• keeping one motion as the “main motion”
• limiting the number of subjects
• using one camera move (not three)
• moving style to the end (1 short style anchor max)

Lock the core seedance prompt (subject + action + camera). Then change only 1 variable per iteration (location OR lighting OR camera move) to improve consistency.