Still Image to Motion — Minimal Prompt Method (Runway Gen-4)

[Use with an input image in Runway Gen-4 image-to-video] The scene comes to life: {{primary_subject_motion}}. The camera {{camera_move}}. In the background, {{secondary_ambient_motion}}. Example fills: - Portrait still → "she blinks slowly and a faint smile forms / slowly dollies in toward her face / curtains stir in a light breeze" - Landscape still → "mist rolls down the valley floor / holds locked and steady / clouds drift right and their shadows crawl across the hills" - Product still → "steam rises gently from the cup / orbits slowly a few degrees clockwise / bokeh lights in the distance shimmer" The rules that make Gen-4 sing: - DESCRIBE MOTION ONLY. The input image already answers what everything looks like; re-describing it ("a beautiful woman with red hair in golden light...") confuses the model and reliably REDUCES motion. If you feel the urge to describe appearance, fix the input image instead. - Three slots, one motion each: subject, camera, background. That's a complete, stable shot. Stack more actions and Gen-4 picks unpredictably which to honor. - Positive phrasing only — Gen-4 does not process negatives. "The camera holds locked and steady" works; "no camera movement" can produce the opposite. - Scale motion words to clip length: "slowly", "gently", "a few degrees" for 5s clips; anything dramatic needs the action to complete within the duration or it warps. - Iterate by changing ONE slot at a time; regenerating with an identical prompt already gives useful variety.