Prompt Generator
Pick your options, copy the prompt, paste it into Gemini or ChatGPT (Step 1) or an image-to-video model like Kling or Veo (Step 2). Every prompt keeps your face unchanged.
Your prompt
Get the style you asked for: start a fresh chat in Gemini or ChatGPT (old edits in the same chat leak into new results), and match your photo to the style — makeover looks (Clean girl, Headshot) need a close-up where your face fills the frame; full-body shots with scenery work best for Polaroid, Figurine and cinematic styles.
Only edit photos of yourself or people who gave you permission — and label AI edits when you post them.
What this actually does
The generator writes a prompt. It does not touch your photo — nothing is uploaded, nothing is processed here, and no image ever leaves your browser. What you copy is plain text that you paste into whichever tool you already use, along with your own photo. That is also why it is free and needs no account.
When to use this instead of a library prompt
The prompt library is the better starting point when something in it already matches what you want. Every entry there has been run, and each one carries a note recording what went wrong in our own result — a hand that lost its fingers, a date stamp that never rendered, a face that drifted by the last frame of a clip. That is knowledge you cannot get from a prompt you have only just assembled.
Use the generator when the combination you want is not in the library: a Y2K digicam photo of you and your dog in moody low light, or a half-orbit on a portrait lit by candles rather than sunset. You trade tested phrasing for a shape nobody has written down yet, which usually costs one or two extra runs.
How the parts fit together
A Step 1 prompt is assembled in a fixed order: who is in the photo, the style you picked, an identity clause, then mood and any extras, and a closing line asking for a result that looks authentic rather than artificially perfect. The identity clause changes with the subject — a group photo asks for everyone's features to be kept, and a pet photo asks for its markings as well as your face.
A Step 2 prompt is motion, then an instruction to hold every face across frames, then the camera move, the vibe, and a five-second duration. Short is deliberate: drift compounds frame by frame, so the end of a clip is almost always worse than the start.
Four things worth knowing before you paste
- One strong idea beats a stack. Extras compound. Grain plus a light leak plus a flare plus soft focus can wash a photo out until there is no contrast left in the face. Add one, look at it, then decide about the next.
- An identity clause reduces drift, it does not remove it. Every prompt built here carries one, and faces still move. Compare the result with your original at full size rather than trusting the clause.
- "Static camera" is a hint, not a lock. In our own tests the video models pushed in anyway. If a steady frame matters, expect to trim the clip.
- Match the photo to the style. Close-up styles need a close-up. The full checklist is in what makes a good source photo.