AI UGC videos: a person, your product and a voice

Models for UGC videos

What UGC ads are and why they work

UGC stands for user-generated content. In advertising the term covers clips that look as if a real person recorded them: a phone held out in one hand, ordinary home lighting, the product in frame and someone talking in the first person. In a TikTok, Reels or Shorts feed, a clip like that reads as a post rather than a banner.

The format runs on simplicity: a person names a familiar problem and shows how the product handles it. AI removes the expensive part of the job, so there is no casting, no lighting setup and no reshoots. You can make several UGC creatives with different faces, hooks and backgrounds and keep the one your audience responds to best.

How to make a UGC video with your product

1. 1

   Write the script

   Hook in the first line, then the problem, the product and a call to action. Write for the ear: short sentences and everyday words.

2. 2

   Voice the text

   In Voiceover, pick a voice that suits your character. The length of the finished track tells you what duration to set for the video.

3. 3

   Prepare the frame

   You need sharp photos of the person and the product. If you have no suitable shot, make one in the Photo section, for example someone holding your package.

4. 4

   Generate the clip

   In Video, add the photo as the first frame or as references, then choose 9:16 and a model. Describe the gestures, the eye contact and what the person does with the product.

5. 5

   Sync the voice to the face

   In Lip sync, upload the clip and the track. If the audio runs longer or shorter than the video, pick a length fitting mode.

6. 6

   Make variants

   Change the hook, the face or the background and run it again. A few versions of one ad are easier to test than trying to guess right the first time.

How to write the script and the hook

Viewers decide whether to keep watching before they even notice it is an ad. So the first line and the first frame have to hook them: a question, a familiar pain, a surprising result or something happening with the product that people want to see through to the end.

After that, stick to a simple order: the problem, the product, how to use it, what changed, and the call to action. One idea per clip. Speak in the first person and in plain words, the way you would text a friend. Read the script aloud with a stopwatch, and if it does not fit the clip, cut words instead of speeding up the voice.

Hooks for the first seconds

• A question the viewer silently answers yes to: "Do you keep losing your keys in your bag too?"
• The result in the first frame: show the outcome up front and explain how you got there later.
• A familiar scene: morning, the kitchen, a traffic jam, a moment viewers recognize from their own day.
• The product in use from the very first second: unboxing, the first press, the first sip.
• Direct address: the person looks into the camera and says right away who the video is for.

How to choose a model

If you have one good photo where the person is already holding the product, a model that starts from a first frame is enough: Kling 3.0 and Veo need nothing else. To end on a planned shot, such as a close-up of the product, add a last frame as well. Kling 3.0, Kling 3.0 Omni, Seedance and Veo 3.1 Fast and Quality support that.

When the person and the product are in separate photos, use a model that takes references: Kling 3.0 Omni, Seedance, WAN 3.0 and MiniMax H3 accept several images at once. Seedance and WAN also take an audio reference, and MiniMax H3 picks its mode on its own based on what you attach.

Length matters too. Veo makes short clips of one fixed length, which is enough for a hook or a single scene. For a longer story, go with Seedance 2.5 or WAN 3.0, which make the longest clips of any model in the table. The same table lists exact durations and resolutions.

Talking photo and your own AI UGC creator

The shortest route is a photo plus a finished track. In Lip sync, InfiniteTalk turns a single picture into a 720p video of that person speaking along with your audio, so you do not need to generate a clip in Video first.

If you want the same creator across a series, set them up in the Characters section: it holds photos of one person, and you pick the character in the video composer before a run. That makes it easier to keep one face from clip to clip. For gestures that feel real, film yourself on your phone and let Kling Motion Control copy your movement onto the character.

Whose face can be in the frame

• Upload a photo of a real person only with their consent, preferably in writing: they should know their face and voice will appear in an ad.
• Do not use celebrities, influencers or random people from social media, not even for a joke.
• Clone only your own voice or the voice of someone who has agreed to it.
• Do not pass off a generated character as a real customer or credit them with experience they never had.
• Ad platform rules on AI-made content keep changing, so before launch check what labels your platform requires and what it allows for your product category.

More tools for UGC ads

• AI voiceover
• Lip sync
• ElevenLabs
• Kling Motion Control
• Nano Banana
• Product ad video
• Videos for TikTok and Reels
• Pricing