Text to Video: How to Write a Clear AI Video Prompt

Start with one shot

A useful text-to-video prompt describes a scene that can happen within a single short clip. Choose a subject and one main action before adding a visual style. Asking for several locations, costume changes, and cuts at once makes the result harder to control.

Use four concrete details

Write down the subject, action, setting, and camera movement. Add lighting when it matters to the scene. Instead of “make an amazing coffee ad,” try:

Close-up of espresso pouring into a clear glass on a wooden table. Steam rises in soft morning window light. The camera slowly pushes forward. Keep the glass centered.

The prompt tells the model what moves and what stays steady. “Cinematic” can be a useful style cue, but it does not replace those instructions.

Choose settings before generating

Open the creation workspace, enter your prompt, and review the available model, aspect ratio, duration, and credit cost. A vertical composition suits a phone screen; a horizontal composition gives a landscape more room. Available options vary by model.

Review one thing at a time

Watch the result for subject consistency, motion, framing, and unwanted details. If the camera moves too quickly, ask for a slower move while keeping the subject and lighting. Changing one instruction at a time makes it easier to understand what improved.

For a longer sequence, plan separate shots and generate each one individually. Generating a clip does not guarantee perfect continuity between shots, so review each transition before assembling your sequence.

When to start with an image

If you already have a product photo or a specific composition, an image can provide a more concrete starting point. Read the image-to-video guide, start creating, or compare plans.