Runway’s video models respond best to prompts that lead with motion: how the camera moves, then what the scene is, then the details that matter. Clear, direct sentences beat long lists of adjectives.
Runway doesn’t use negative prompts — writing “no blur” can even add blur — so the generator phrases exclusions positively, for example “smooth, stable motion” instead of “no jitter”.
Opening with the movement (“Slow push-in:” or “Tracking shot:”) tells Runway how the shot should feel before anything else.
Say what you want to see. Positive phrasing avoids accidentally adding the thing you wanted to remove.
When you start from an image, the frame already sets the look — keep the prompt about the movement and the action.
A single clear action reads better than several events squeezed into a few seconds.
Runway doesn’t support negative prompts, so the generator turns common exclusions into positive statements inside the prompt.
The structure works for Runway’s current Gen models; set the duration and aspect ratio in Runway itself.