Prepare the image before the motion prompt
Select an image-to-video mode that accepts a start image. Inspect the image before uploading: the face should be readable, the hands needed for the action should be visible, and the frame should include room for the movement. If you want a suited character in a Paris apartment, establish those features in the image rather than asking the video prompt to replace the subject and background at the same time.
Example portrait description: Fictional adult man, honey-blond bob, curled handlebar moustache, brown houndstooth suit and tie, serious expression. Three-quarter body portrait in a softly lit apartment, both hands in a relaxed boxing guard, realistic proportions, empty space around the fists, vertical composition.
Copy a motion prompt for a restrained shadowboxing shot
The man keeps a blank, serious expression. He raises his fists into a relaxed guard, throws one short jab and returns his hand to guard, then pauses and looks at the camera. Small natural weight shift. Preserve his face, bob haircut, moustache, suit and apartment. Static camera, one continuous shot, no cuts.
Put the performance before the preservation details. The image already establishes the appearance, so the main job of this prompt is to say what happens next. If the result changes too much, simplify the action first. Repeating the same identity sentence many times is less informative than making the required movement smaller and clearer.
Change the action without rewriting the character
Slow look: The man turns his head slightly toward the camera, holds eye contact for a moment and gives a very small bow. His expression remains serious. Preserve the subject, outfit and room. Static camera, one continuous shot.
Small dance: The man makes one gentle shoulder sway and a short side step, then returns to his starting position. His expression remains completely deadpan. Preserve the face, hair, moustache and suit. Fixed camera, restrained movement, no cuts.
Keep the input image, model, duration and framing unchanged when comparing these actions. Change one instruction at a time so you can tell whether the problem comes from the motion request or from a different starting image.
Choose duration and sound outside the prompt
Controls vary by model and platform. Higgsfield’s official Kling 2.6 Pro image-to-video reference documents an image input, a text prompt, selectable five- or ten-second duration, and a separate sound control. Those are settings for that specific endpoint, not universal limits or free-tier entitlements for every Kling product. Check the documented Kling 2.6 Pro settings.
Start with a duration that gives one gesture time to happen without forcing a long sequence. Check the renderer’s actual quote before creating extra variations. If you need silence, use the sound control when the model provides one rather than relying only on the words “no audio” in the prompt.
Use motion control for a specific recorded performance
Kling image-to-video and Kling Motion Control solve different tasks. Image-to-video uses a starting image and text to invent movement. Motion Control uses a source image and a provided video to guide a performance. A copied jab prompt does not substitute for a reference clip when the exact rhythm of a dance or boxing routine is the part you want to preserve. Compare the documented motion-control inputs.
Generate here or take the prepared prompts elsewhere
On jeanphilai.app, Create your own → Start with an idea → Review & continue prepares a portrait prompt and a motion prompt you can copy without signup. Rendering the new portrait costs $1.00, followed by $18.90 for the five-second silent HD 720p video after portrait approval. For a longer predefined performance, the featured 19-second Apartment shadowboxing template costs $9.90 and uses an uploaded photo directly. Prepare your own prompts. Review the current rendering prices.