Free AI video prompt generator: UGC that looks filmed on a phone

AI video models are trained on huge amounts of polished footage, so when a prompt is vague they fill the gaps with what they have seen most: smooth camera moves, perfect light, slow motion and a person who presents the product to the lens. That is the look of a TV ad, and on TikTok or Reels it is the look people scroll past.

UGC is the opposite: a phone held at arm's length, a bathroom with a towel on the rail, the light that happens to be on, a quick gesture done a hundred times before. The prompts this tool writes describe exactly that, in the order video models follow best, and they leave out everything that pulls the model back towards cinema.

1. 1Lock the references. "Keep the exact product from image 1 in every frame, same shape, same size, same colours." Add "Keep the reference face from image 2" when you also attach a person.
2. 2Say it is phone footage. "Real amateur iPhone footage", then who holds the phone and from where, then "no filter". This one line moves the model away from its cinematic default more than any adjective.
3. 3Set the scene in three to five words. "Small bathroom, toothbrush cup, folded towel." Two to four ordinary props make a place feel lived in. Long descriptions of the room make results worse, not better.
4. 4Name the light and where it comes from. Harsh overhead bathroom light, morning light through curtains, a phone flash in a dim room. Real light has a direction and casts shadows; "beautiful lighting" does not.
5. 5Start the action in the first frame. Use "already": "already pressing the dropper onto her fingertips". Without it many models spend the first second on an intro and you pay for footage you cut.
6. 6One light detail on the skin or the product. "The overhead light catches the fine pores on her cheekbone." A single precise texture is what makes the clip read as real.
7. 7Close with the raw-footage line. "Completely raw unedited iPhone footage, light bounce on the white sink, no filter, no color grading, film grain, one continuous recording."

A serum filmed as a morning bathroom selfie. Notice what is not there: no adjectives about the product, no "beautiful", no camera moves, no second action.

Keep the exact product from image 1 in every frame, same shape, same size, same colours. Real amateur iPhone footage, phone held in hand by the woman, close selfie angle, no filter. Small bathroom, toothbrush cup, folded towel, fogged mirror. Harsh overhead bathroom light from above. Woman in her late twenties in an oversized grey t-shirt, hair in a loose bun, already pressing the dropper of the product onto her fingertips and patting it into her cheek, quick natural movements, glancing between the phone screen and the mirror. The overhead light catches the fine pores and the slight shine on her cheekbone. Completely raw unedited iPhone footage, light bounce on the white sink, no filter, no color grading, film grain, one continuous recording.

Negative prompt: cinematic, studio lighting, color grading, slow motion, text overlay, subtitles, watermark, extra fingers, deformed hands, warped product, changing label

• Timestamps or several actions in one clip: the model blends them or ignores the order. One action per clip, then edit clips together.
• Describing the product's design in words: the reference image does that better, and words make the model redraw it. Call it "the product".
• Sentences that sound like dialogue ("this changed my mornings"): video models with sound often speak them out loud. Describe physical actions only, and put real dialogue in quotes on purpose.
• "Slowly", "steady", "gracefully": real phone videos are quick and slightly messy. Slow words produce slow-motion, ad-looking footage.
• Having the person present the product to the camera: a natural everyday gesture (applying, pouring, zipping the bag) reads as UGC; holding it up to the lens reads as an ad.
• Cinematic places (offices with city views, lofts, studios): kitchens, bathrooms, cars, couches and desks look like real customers' homes.

Pick the camera the way a customer would film themselves. A selfie suits talking and applying; a phone propped on a counter suits cooking, cleaning and anything that needs two hands; a mirror selfie suits outfits and accessories; POV suits unboxing and gadgets; filmed by a friend suits outdoor and sport.

• Bright daylight: "bright daylight from a window"
• Morning light: "soft morning light through curtains"
• Golden hour: "warm golden hour sunlight from a side window"
• Evening lamp: "warm lamp light in the evening"
• Bathroom overhead: "harsh overhead bathroom light"
• Kitchen fluorescent: "cool fluorescent kitchen light"
• Phone flash: "phone flash in a dim room"
• In the car: "daylight through a car windshield"

Write the script first with the UGC script generator: its shot list gives you one line per clip. Create a consistent person with the AI influencer prompt generator and use the image as the face reference. Then generate each clip with your product photo, and compare models on the AI video models page: Seedance 2.5, Kling 3.0, VEO 3.1 and Wan 3.0 each have their own prices and limits.