HiAPI
Showcases

Kling 3.0 Motion Control Prompt Formula Card: A Structured Shot Description Demo

Use a real motion-control task to demonstrate how to write a "Prompt Formula Card": the character appearance comes from a reference image, while the action and expressions are driven entirely by a reference video. The prompt only supplies the scene, style, and cinematic language, and does not describe the action itself. Composition of the work: 1. Character reference image (image_url): front-facing half-body, with the head, shoulders, and torso fully clear, satisfying the 2:5–5:2 aspect ratio constraint. 2. Action reference video (video_url): a single continuous shot, with the subject's head and torso always visible, and a duration within the 3–30 second range. 3. Prompt: written in a three-part structure of "subject styling → environment and lighting → camera and texture". Descriptions of action and expression are deliberately left blank and left to the reference video to determine. 4. Parameters: resolution=720p; character_orientation=video makes the orientation follow the reference video, suitable for complex actions and longer durations; keep_original_sound=true preserves the original audio of the reference video. Use cases: show developers and creators the input boundaries of this pipeline—which information should be written into the prompt and which must be carried by the assets; it is also suitable as a templated starting point for action-transfer short videos, where the same prompt structure can be reused by simply replacing the reference image or reference video.

Style Motion Transfer · Prompt Engineering · Storyboard

Model Kling 3.0

Reference inputs

  • image_url

    image_url

  • video_url

Prompt

Try this model
A medium shot of the character in the reference image, standing in a modern photography studio with soft lighting, featuring a large diffused window as the key light and a faint cool-toned rim light coming from behind. Always ensure the subject's head, shoulders, and torso remain fully visible and centered in the frame, with clean negative space at the top of the image. Cinematic color grading, shallow depth of field, natural skin texture, fine film grain, 35mm equivalent lens. The camera is fixed and pushes in very slowly; posture, expression, and pacing follow the reference video entirely, and the prompt does not describe any additional action.

Copy the full prompt and adapt it for your own generation.

Tags