NeuralBox

How to write a video prompt

A video model renders what you describe. A vague prompt leaves the rest to chance. A good prompt saves runs and money.

The five-part formula

Describe, in order: who is in frame, what they do, where, how the camera moves and the light. For Veo 3.1 and Kling 4.0, add sound.

Who
"An old fisherman in a yellow raincoat."
Action
One main action: "pulls a net out of the water."
Where
"A wooden pier at dawn."
Camera
"Slow push-in", "side view", "drone shot."
Light
"Soft morning light", "neon", "rim light."
Sound
For Veo and Kling 4.0: "gulls, waves". Put spoken lines in quotes.

Before and after

BeforeAfter
cat playingA ginger kitten chases a ball of yarn across a wooden floor, camera at floor level, warm evening light
car drivingA red convertible drives a mountain road, drone shot, sunset, long shadows

Common mistakes

Too many actions in a short scene: five seconds fit one main action. Describing what is already in the photo when animating it: describe only the motion. Contradictions like "night" and "bright sun".

Try your prompt

Every video model on one balance.

Make a video