رجل بمعطف أسود يفرّ عبر حشد يطارده

What this shot looks like

A man in a black coat sprints from a chasing crowd. The camera tracks behind him, then switches to a side-tracking move as he knocks over a roadside fruit stand mid-stride, picks himself up, and runs on. Panicked voices and the clatter of spilled produce fill the soundscape.

It's a compact action-film grammar exercise: protagonist flight, environmental destruction, and a camera-angle switch — all in one prompt.

When to use it

This is a strong evaluation prompt for AI video models:

• Chase-sequence test — does the model hold subject continuity (same coat, same silhouette) across the camera cut?
• Physical-object interaction — knocking over a fruit stand requires fruit, stand, momentum, and a recovery beat
• Directed sound test — many models still under-deliver on prompted ambience; this one explicitly asks for crowd panic noise on top of the action

Tips for customizing

• Swap "black coat" with another wardrobe call (red hoodie, white shirt) to test how rigidly the model honours costume detail
• Replace "fruit stand" with a different obstacle (newsstand, bicycle, café table) — same beat, different physics
• Add a city tag ("Hong Kong wet market", "Tokyo back alley") to test cultural set-dressing

Related

• More cinematic video prompts
• See all video models we've tested