In this article
A cinematic direction begins with a decision about the scene, rather than a long list of impressive adjectives. What should the viewer notice first? What should change before the shot ends? You can answer both questions before choosing a video model.
This exercise uses an imaginary lakeside scene to turn a broad intention into a useful creative brief. It is a planning method, not a claim that every model will reproduce every instruction.
Start with the light
Choose a time of day and a visible light source. For our scene, try low evening sunlight arriving from the right, with the woman's face turned slightly toward it. Decide whether you want to see her expression or read her silhouette. Those are different pictures, even if the location stays the same.
Write the lighting idea in one sentence. Avoid asking for bright midday sun, deep night shadows and sunset colors at the same time unless that conflict is intentional. A useful starting point is: soft evening light across the face, cooler mountains in the distance.
Give the frame a clear subject
Choose the shot size before adding scenery. A wide view can make the lake and mountains part of the story. A closer view makes the woman's reaction the subject. Neither choice is automatically better. Pick the one that serves your idea.
For this exercise, place her toward the left of a wide frame and leave space in the direction she is looking. Keep the red coat as a deliberate point of attention. Describe the surroundings briefly so that background details do not compete with the action you care about.
Describe one movement
Separate the subject's action from the camera's movement. The woman might turn toward the lake while the camera remains still. Alternatively, she might stand still while the camera moves slowly closer. Choose one version for your first attempt.
A complicated combination of running, turning, zooming and circling makes it harder to identify which instruction needs revision. Start with a small action that has a clear beginning and ending. Add complexity after you have a result worth building on.