AI Video Prompt Generator

Camera Movement & Lighting for Runway, Sora & Pika

Note: Runway's current flagship model is Gen-4 (Gen-2 is outdated). Across Runway, Sora, and Pika, standard cinematography terms (pan, tilt, dolly, orbit, crane) work reliably — but one camera move per clip is the rule that matters most: combining moves like "zoom in while panning left and tilting down" tends to confuse the model into producing muddled motion. This tool deliberately only lets you pick one movement at a time.
Scene
What's happening in the scene — the core action or subject.
"Slow dolly in" behaves differently from an unqualified "dolly in" — leaving this blank lets the model pick a default.
5-8s is most reliable; longer takes tend to degrade.
Video Prompt
About
Platform Notes
Contact

About the Video Prompt Generator

This tool builds a text-to-video prompt using standard cinematography vocabulary that works consistently across Runway, Sora, and Pika: subject and action, a single camera movement, lighting, lens/style, duration, and aspect ratio. It deliberately restricts you to one camera movement at a time — research and platform documentation both point to compound instructions ("zoom in while panning and tilting") producing confused, muddled motion rather than the intended effect. Everything is generated locally in your browser.