Write video prompts with clearer motion and fewer surprises
A useful video prompt describes one subject, one main action, the setting and the camera. Keep movement modest when preserving a face, product shape or full-body pose matters.
Use a simple structure
Subject + setting + action + camera + lighting. Example: “A ceramic cup on a wooden table, steam rising gently, locked camera, morning window light.”
For image to video, focus on the motion you want added to the existing picture. For text to video, also describe the subject and environment because there is no reference image.
Three starting prompts
Portrait: “The person slowly looks toward the lens and smiles softly. Gentle breathing, steady camera, consistent daylight.”
Product: “A glass perfume bottle rests on stone. Light glides slowly across the glass. The bottle stays still, close-up shot.”
Landscape: “A calm coastal scene at sunrise. Small waves move toward shore and thin clouds drift slowly. Static wide shot.”
When people look distorted
Reduce the number of people, the speed of movement and the complexity of hand gestures. Keep the full body in frame for a dance shot, avoid crossing limbs and test a small movement first.
For image input, use a clear source with visible anatomy and preserve its orientation. Prompt wording and framing can help, but they cannot eliminate model limitations or guarantee an artifact-free result.
Begin with a five-second take.
Photo or text. Landscape or portrait. Check the current price and node availability in the studio.
Start creating ↗