How to make a Jean Phil AI video (step by step)

To make a Jean Phil video, turn one photo into a portrait with his look (thick blond bob, blunt fringe, curled handlebar mustache, houndstooth suit, straight face) using an image-edit model, then animate that portrait with an image-to-video model and a 10-second, front-camera, deadpan scene prompt. Jean Phil AI does both steps from one photo.

Step by step

1. Pick one clear photo Front-facing, good light, one person or one pet, no sunglasses.
2. Make the portrait with the look An image-edit model (Seedream, Nano Banana, GPT Image) with your photo and a prompt for the look: helmet bob, curled mustache, long neck, suit, chin up, deadpan.
3. Animate it Image-to-video model (Seedance 2.5, Kling, Veo) with the portrait as reference image 1 and a scene prompt: front camera, real place, deadpan, ~10 seconds, 9:16.
4. Post it like he does No music or one trending sound, a short caption ending in “Oui Madame.”, posted from your phone.

The formula that makes it read as Jean Phil

• The bob. A thick, glossy honey-blond helmet: a rounded dome on top, a heavy square fringe cut dead straight just above the eyes, straight sides that bell out wide at the jaw. Black on the yellow-polo version.
• The mustache. The same blond as the hair, long thin arms that end in tight round curls. Together with the bob it is the whole silhouette.
• The neck. Strikingly long and thin, on narrow sloping shoulders, with a slim, gaunt, pale face — it makes the bob look even bigger.
• The suit. Brown houndstooth three-piece, white shirt, navy-and-red striped tie. Copycats swap in pink houndstooth or a yellow polo.
• The face. Long and narrow, ruddy and freckled, eyebrows so pale they almost vanish, eyelids half-closed. He raises his chin and looks down his nose into the front camera — never a smile.
• The clip. About 10 seconds, vertical, front camera at arm's length, real places, no dialogue, a caption ending in “Oui Madame.”

Starter prompts

Turn the person in this photo into Jean Phil: a thick blond bob with a straight fringe, a blond handlebar mustache with curled ends, a brown houndstooth jacket and a striped tie, completely straight face. Keep their own face.

Vertical selfie video, 10 seconds: the man from the reference image shadowboxes on a Paris street, completely deadpan, then walks up and stares into the camera.

These are simplified starters; results vary a lot by model. Jean Phil AI's own pipeline does both steps from one photo.

What we saw making our samples

• One selfie, three looks, same face. From a single AI-generated selfie we produced the original, pink-suit and yellow-polo looks, and the face was still clearly the same person in every one.
• Colour and neck make it Jean Phil. Our first portraits had a dark mustache under blond hair and an ordinary neck — they looked like a costume. Matching the mustache to the hair and giving him the long, thin neck is what made it read as Jean Phil.
• The curl is the signature. The upturned, curled tips of the mustache are what make the silhouette recognisable even at thumbnail size.

The other way: video-to-video

The originals are widely reported to be real phone clips with the character swapped in. If you want your own exact movement — your walk, your dance — film it first, then run it through a model with a character-reference or motion-transfer mode, using your Jean Phil portrait as the reference. It's slower and pickier about the source clip, but the motion is real.

Questions