wowprompt

Three AI photo prompt tips talking head at a desk by a window

A ten second vertical talking-head clip: a young woman creator sits at a pale wood desk by a window in her Kochi flat, counts three tips for better AI photo prompts on her fingers and tells the camera to name the light, name the lens and say what to leave out. Cool clean light, true colour, made for a reel or a short. Works with your own photo.

A young woman in a sage linen shirt at a wooden desk counting on her fingers as she talks to the camera beside a window
The output this prompt produced, unedited.
Works with or without your photo.

Optional. Attach a clear, front-facing photo to keep your own face. Without one, the prompt creates an original person.

The prompt

Logline

One creator, one desk, three quick tips counted on her fingers, said straight to the lens.

Identity

With a photo attached: the person in the photo is the subject. Keep their face exactly as it is, with bone structure, eye shape and spacing, nose, lips, jawline, hairline, skin tone and apparent age all unchanged in every frame, and keep their own body: build, height and proportions, shoulders, arms, hands and legs, with the same skin tone on face, neck, arms and hands. Use the photo only for who the person is. Ignore its pose, expression, clothing, hairstyle, background, framing and light, and rebuild all of them exactly as described below, even when the photo shows them standing straight or dressed differently.

Subject

Framed from the waist up, she sits square to the lens, forearms on the desk. Seconds 0 to 2: she leans in, eyebrows up, raises her right index finger. Seconds 2 to 8: she adds a second finger, then a third, nodding on each, her left hand open on the desk. Seconds 8 to 10: she drops the hand and smiles. She says: "Three things. Name the light. Name the lens. And say what you don't want." Lips match the words. A sage green linen shirt with the sleeves rolled, small gold studs, kajal.

Set

A pale wood desk with a closed laptop showing no marks, a notebook and a small potted pothos. Behind her a white wall and the edge of a window with coconut palms soft outside.

Lens

A 50mm look at f/2 at eye level, locked off with a slight handheld breath.

Grade

Cool soft daylight from the window on the left, true colour, real skin with pores. Sound: her voice clear and close, faint birdsong, no music.

Negative prompt

airbrushed skin, plastic skin, waxy face, changed face, morphing face, face changing between frames, lip sync drift, distorted hands, extra fingers, wrong finger count, laptop logo, screen text, orange cast, yellow tint, golden glow, warm wash, subtitles, text, logo, watermark

Thumbnail prompt

For the cover image of this video. Paste it into an image model such as GPT Image.

Frame

A tall cover frame at a pale wood desk by a window. A young woman in a sage green linen shirt sits in the lower half, leaning toward the lens, holding up three fingers of her right hand beside her face, eyebrows raised, a confident smile. A closed laptop and a potted pothos soft beside her, palms beyond the window. Cool soft daylight, true colour, real skin. The top third is a clear white wall for the words.

Identity

With a photo attached: keep the face from the photo exactly as it is. Without a photo: an original Malayali woman of about twenty-seven with deep brown skin, curly black hair to the shoulders and a small scar on her forehead. Do not base the face on a real person.

Text

  • "3 AI PHOTO"
  • "PROMPT TIPS"

Heavy black condensed capitals on the white wall, "3 AI PHOTO" with the 3 twice the size, stacked in the top third, clear of her face, readable at small size. Nothing else is written anywhere. Spell each line exactly.

ai photo promptsprompt writing tipstalking head reelai image tipscreator videovertical talking head

How to run this

Copy the prompt and paste it into Veo, then set the aspect ratio to 9:16. Upload your reference photo before pasting, not after: a model that has already started without it will invent a face instead of keeping yours.

The prompt is written in Markdown with each part under its own heading, which is what the copy action hands over. Change one section at a time, keeping the rest intact, so you can tell which change did the work.

Related prompts

All video