An original portrait prompt to start with
A fictional tall, slender adult man with a blunt honey-blond bob haircut, a small curled handlebar moustache and an elongated neck. He wears a fitted black-and-ivory houndstooth suit over a dark shirt. Deadpan expression, relaxed shadowboxing stance, both hands visible. Full-body fashion editorial on a quiet Paris terrace, soft natural afternoon light, muted warm tones, realistic proportions and textured wool. Static eye-level camera, enough room around his hands and feet. No text, logos or watermark.
An original shadowboxing motion prompt
Animate the approved portrait as a five-second continuous shot. The man performs restrained, playful shadowboxing: one short jab, returns to guard, then a gentle weight shift. Keep his face completely deadpan. Preserve the bob haircut, curled moustache, houndstooth suit, background and camera position. Natural hand anatomy, small controlled movements, no cuts, no additional people, no captions, no audio.
Customize the scene without losing the character
Change one variable at a time: the terrace becomes a quiet metro platform, or afternoon light becomes a softly lit studio. Preserve the facial description and outfit if character consistency is your goal.
For a runway version, replace punches with two composed walking steps. For a disco version, use a gentle shoulder sway and a small side step. Large motions and moving cameras increase the chance of inconsistent anatomy.
A prompt is guidance, not an exact motion transfer
Image-to-video generation interprets your text and starting image. It does not guarantee the timing or choreography of an existing Jean Phil clip.
The studio supports a personal photo as an appearance reference and a 3–30 second clip as a motion reference when live generation is configured. The photo is used to create an approved portrait first; motion control then uses that portrait and your clip. Identity and choreography are model-guided, so review the output rather than expecting an exact copy.