2 reusable prompts
Cinematic AI Video Prompts
Cinematic scenes land when the world feels coherent before the impossible detail appears. Start with light, scale, and camera intent.
Prompt collection
Pick a shot plan and make it your own
Cinematic
The Floating Library
Imaginative environmental motion with strong visual identity
One continuous 5-second magical-realist shot inside an immense old library flooded with still ankle-deep water. Hundreds of open books float gently in the air beneath a vaulted glass ceiling, and one luminous white koi swims through the air between them. From 0-3s, the camera glides forward just above the reflective water on a 28mm lens while the koi crosses from right to left and loose pages respond to its movement. From 3-5s, the camera tilts upward as sunlight breaks through the ceiling and the floating books slowly align into a spiral. Coherent large-scale architecture, elegant restrained magic, realistic water reflections, warm dust-filled light, stable book shapes, graceful motion. Audio: soft page rustle, water movement, one distant resonant bell. No cuts, no people, no readable text, no logos, no watermark.
Cinematic
The Lost Greenhouse
Atmospheric narrative shot with clear camera choreography
One continuous 5-second cinematic science-fiction shot inside an abandoned glass greenhouse at dawn. An adult explorer in a weathered ivory pressure suit steps through waist-high silver grass toward a dormant circular machine covered in vines. From 0-2s, the camera follows low behind the explorer on a 28mm lens as dew bends under each step. From 2-4s, the machine awakens with a warm amber ring of light and thousands of dust particles rise into the sunbeams. From 4-5s, the camera arcs gently to reveal the explorer's astonished profile and the glowing machine in the same frame. Photorealistic production design, coherent scale, stable suit details, volumetric dawn light, subtle handheld weight without shake. Audio: soft servo activation, grass movement, distant birds, low atmospheric swell. No cuts, no dialogue, no text, no logo, no watermark.