AI UGC Video Generator

UGC, short for user-generated content, is the style of ad that looks like a person filmed it on their phone: a creator at a bathroom counter or in a kitchen, holding the product up to the camera, talking the way people talk. Brands use it because it reads as a recommendation rather than a commercial. An AI UGC video generator makes that clip without the creator, the shipping or the reshoot.

On Mage the whole clip comes from one model. Cherry 2 Pro generates the picture and the voice together, so the creator says your lines word for word while holding your product, which it takes from your photo as a reference. You can describe the creator in a sentence or cast a saved character with their own voice, so the same person fronts a whole series of ads.

This page is about product ads with a scripted presenter. A virtual person who posts a feed of their own over months is an AI influencer, and a portrait that speaks a recording you already have is a talking photo. Both have their own guides, linked further down this page.

People speak about two and a half words a second on camera, so budget the script by the clip: 20 to 25 words for 10 seconds, 35 to 40 for 15 seconds, and 50 to 75 for a 20 to 30 second clip. Leave half a second of quiet at each end. A script that runs long gets rushed or cut off, which is the most common reason a UGC clip feels off.

Spend the first two seconds on the hook. A few patterns work across categories: surprise ("Okay, I didn’t expect this to actually work"), a callout ("If your skin gets dry by noon, stop scrolling"), a confession ("I’ve been gatekeeping this for months"), a question, or an unboxing ("It’s here. Let’s open it."). Follow it with one concrete detail shown on camera, and close with a call to action of a few words.

Write casually. "Honestly," "okay" and small pauses read as real, while marketing phrases read as an ad. Every claim about the product, how it feels, smells, fits or works, should come from your own product facts. A presenter can describe what the product is; they should never claim results, awards or reviews it does not have.

• Budget about 2.5 spoken words per second.
• Put the hook in the first two seconds.
• One gesture per beat: hold up, tap, open, apply, turn.
• End on a call to action of a few words.
• Write two hooks for every script and test them.

Pick the format before you write. A review goes hook, what it is, one standout detail on camera, verdict, call to action. An unboxing goes box in hand, opening reaction, product reveal, first impression. A how-to walks through two or three steps of use and shows the result. A try-on goes from before, to putting it on, to a turn that shows the fit. A routine places the product in a morning or evening routine and says why it stays. A problem–solution opens on a relatable problem, brings in the product as the fix, and shows the proof on camera.

The setting does half the work. Real rooms read as UGC: a bathroom counter, a kitchen, a bedroom, a parked car. Ask for a handheld smartphone front camera at arm’s length, natural window light and a lived-in background, and add "not a studio ad." Then add "no music": UGC usually runs on voice and room tone.

Describe the creator in one line, their age range, look and energy, or cast one you have saved. A saved character keeps the same face across every clip, and when you give the character a voice, an uploaded MP3 or WAV or a short sample made with Seed Audio, Cherry 2 Pro speaks their lines in it automatically. Build one in Characters or with the AI character generator.

A recurring presenter is not the same job as a virtual influencer with a feed of their own; for that, see the AI influencer generator. Use a real person’s face or voice only with their permission: Mage does not allow harmful deepfakes or non-consensual depictions of real people.

Cherry 2 Pro uses images as references rather than as a first frame, so attach the approved opening frame and the product photo, and say which is which. A structure that works: "Handheld vertical smartphone video, UGC style. The woman in the first reference image talks to the camera at her bathroom counter, holds up the bottle from the second reference image, taps the cap and turns it to show the label. She says: … Natural voice, casual delivery, room tone, no music. Keep the product identical to the second reference image: shape, colors, label."

Render drafts at 480p and finals at 720p. 1080p costs more, and most vertical ads are watched on a phone. Lemon makes a cheaper cut: use the opening frame as Lemon’s first frame, turn audio on, and describe the product in words, because a Lemon clip uses either frames or reference images, not both.

Already have a voice line and a portrait? Lipsync animates one image to speak up to 15 seconds of audio, and the AI talking photo guide shows how to write the voice first.

• Say which reference is the creator and which is the product.
• List the actions in the order they happen.
• Quote every spoken line.
• Draft at 480p, render the keeper at 720p.

The free UGC Video skill runs the steps above inside Claude, Cursor or another assistant connected to Mage. You share a link to the product photo and two or three talking points, and pick a format and a creator. The skill writes the script, previews the opening frame with Mango 3 for a few Gems, quotes the video price and waits for a yes, then renders on Cherry 2 Pro, or Lemon for a cheaper cut, and checks that the product matches. Connect Mage and add it from the Mage skills page.

Be open about what the video is. A UGC ad made with AI shows a presenter, not a customer, so do not present it as a genuine review. Platforms such as YouTube ask creators to disclose realistic AI-generated content, and ad platforms have their own labeling rules, so check them before a campaign. For ads beyond the creator format, see the AI ad generator.