UGC Studio: AI Video Ads and Brand Images from One Product Photo

Short-form ad creative has settled on one format: a person holding your product, talking to a phone camera. It converts because it does not look like an ad. The problem has always been production — you need a person, a camera, a product in hand, and a new shoot every time you want to test a different hook.

UGC Studio does both halves of that job. One tab makes the video. The other makes the product photography.

Part 1: UGC video ads

Pick a creator

Twenty AI creators ship with the tool — ten female, ten male, each with a different look and style. Filter by gender, search by name or style, click one.

If none of them fit your market, upload your own photo instead. A single clear, front-facing image is enough; that person becomes the creator for the whole video.

Write the script exactly as it should be said

This is the part people get wrong. The script field is not a brief and not a topic — it is the words that come out of the creator's mouth, verbatim. It is split into clips, not rewritten.

Up to 4,000 characters. Write like a person talks:

Okay so I did not expect the difference to be this obvious. You know that flat, burnt taste supermarket beans have? None of that. And they roast it the week it ships, so the bag is still gassing when it lands on your counter.

Not:

Introducing Northwind Coffee, featuring single-origin beans and a weekly roasting schedule for unparalleled freshness.

The second one reads as an ad the moment it is spoken. The first one does not.

Attach the product

One product photo, under 1.5MB. Clean background helps but is not required — what matters is that the product is clearly visible and in focus. That image is what the creator holds and what the camera lands on.

Format and clip length

• Vertical 9:16 for TikTok, Reels and Shorts. Wide 16:9 for YouTube pre-roll and landing pages.
• 4, 6, 8 or 10 seconds per clip. This is the unit the script is chopped into, not the length of the final video.

The script is split at roughly 2.5 words per second, capped at 8 clips. A 300-word script at 8 seconds per clip becomes 8 clips — about 64 seconds of finished video. The clip count and the exact credit cost update live as you type, so nothing is a surprise.

Pricing

Each clip is one video render, priced the same as the equivalent model in the standard video tool:

A typical 4-clip, 8-second ad is 31,252 credits. Credits are deducted when the job starts and automatically refunded in full if the render fails — you are never charged for a job that did not produce a video.

Lip-sync comes for free

Speech is generated inside the video model rather than dubbed on afterwards. There is no separate voiceover step and no drift between the mouth and the audio, which is the tell that gives away most cheap avatar video.

Part 2: Brand images

The second tab solves the other half of the problem: you have one product photo and you need twenty.

Upload 1-5 reference images of your product — different angles help — then pick what the set should look like:

• Product shots — clean studio packshots on neutral backgrounds. For listings, ads and catalogues.
• Review style — held in hand, phone-camera framing, the look of a real customer photo. This is what pairs with UGC video.
• Brand scenes — styled compositions built around your brand rather than the bare product.
• Lifestyle — the product in use, in a real setting, with context around it.

Add an optional line of art direction ("soft morning light, pale linen and matte ceramic, no people in frame") and choose a format: 3:4, 1:1, 9:16, 16:9 or 4:3.

Generate 2 to 12 images per set at 1,000 credits each. The product stays consistent across every image in the set — that is what the reference photos are for.

Why this matters more than a generic image model

A general image model draws a product. This one draws your product, because it is working from your photos. The label, the shape, the colour and the proportions carry through. That is the difference between images you can ship and images that get you a trademark complaint.

A workflow that actually works

• Generate a review-style image set from your product photos. Pick the best frame.
• Use that frame as the product image in the video tab — the product now looks photographed rather than catalogued.
• Write three different hooks as three separate scripts. Same creator, same product, three renders.
• Run all three. Keep the one that holds attention past three seconds.

Testing hooks is the whole game in short-form advertising, and the cost of a new hook here is one render instead of one shoot.

Common mistakes

Writing a brief instead of a script. The tool does not rewrite you. Whatever you type is what gets said.

Too long. Eight clips is the hard cap. A 1,500-word script does not become a 6-minute video — it becomes a rushed 8 clips. Keep ads under about 400 words.

A bad product photo. Blurry, cropped or badly lit input produces exactly that in the output. Spend the two minutes.

Uploading a huge file. Keep images under 1.5MB. Anything larger is rejected before it reaches the render, and resizing takes seconds.

Corporate voice. "Elevate your daily routine" is not something a person says holding a bag of coffee. Write the way you would text a friend about it.

Availability

UGC Studio is live in the dashboard now — [open it here](/dashboard/ugc-video). Both halves run off the same credit balance as everything else on the platform, starting at $4.99. No subscription, and credits never expire.