How to Make AI UGC Ads for TikTok and Reels with Seedance 2.0

To make an AI UGC ad, write a short prompt describing a creator talking to camera about your product, generate it with Seedance 2.0 on Prospolabs, and you get a clip with spoken audio and matched lip sync in one pass — no actor, no separate voiceover, no lip-sync tool. Here is exactly how to do it, what it costs, and how to spin up dozens of variants for a single launch.

What Seedance 2.0 actually changes for UGC ads

User-generated-content ads — that handheld, lit-by-a-window, talking-straight-to-the-lens look — outperform polished studio spots on TikTok and Reels because they read as a real person, not a brand. The catch has always been supply: every variation meant booking another creator, writing another brief, waiting days, and paying per deliverable. Seedance 2.0 removes that bottleneck. It generates 4 to 15 second clips up to 1080p with native, always-on audio and lip sync, so the spoken testimonial and the mouth movement come out of the same generation.

That single-pass audio is the part most video models still can't do. With Seedance 2.0 you describe the person, the product, the room, and what they say, and the model returns a finished talking clip — room tone included. No bolting a silent video onto a separate lip-sync service, and no timing drift between the voice and the mouth, because they were never two separate files.

For a paid-social team, that collapses the whole production line into one step. The brief, the shoot, the edit, and the voiceover used to be four separate dependencies, each with its own turnaround and its own invoice. Here the prompt is the brief and the generation is everything else. You go from idea to a postable clip in the time it takes the model to render.

Seedance 2.0 runs in three modes, and UGC ads use all three depending on the shot:

• Text-to-video — describe the creator, scene, and dialogue from scratch. Best for fresh hooks where you don't yet have a face or product shot.
• Image-to-video — supply a start frame (and an optional end frame) so the clip animates from a real product photo or a locked creator look.
• Reference-to-video — cite up to 9 images, 3 videos, and 3 audio clips with @Image1 / @Video1 / @Audio1 syntax to keep a face, product, or voice consistent across a whole campaign.

How to make a UGC ad on Prospolabs, step by step

1. Write the hook as a prompt

UGC lives or dies on the first second. Write the prompt the way you'd brief a creator: who they are, where they are, how the camera behaves, and the exact line they say. Something like: "A relaxed twenty-something on a sofa in soft daylight looks into the camera and says enthusiastically, 'Okay I did not expect this to actually work' while holding up a cosmetic jar, slight handheld sway, warm room tone, no on-screen text." Specifying the spoken line is what drives the audio and lip sync.

2. Pick the model and tier

For drafting hooks and high-volume variants, start with Seedance 2.0 Fast for cheap variants at 480p or 720p. Once a hook proves out, re-run the winner on Seedance 2.0 at 720p or 1080p for the version that actually goes live. Same prompt, higher fidelity.

3. Set length and aspect, then generate

Keep clips in the 4 to 6 second range — UGC ads are punchy, and shorter clips cost less per generation. Generate from the playground or hit the API directly. A call is a `POST /v1/generate` with model id `seedance-2` or `seedance-2-fast`; it returns a `generation_id`, and because the job is async you poll until an `output_url` comes back (valid for 7 days). The playground and the API charge the exact same price, so prototype wherever you like.

What it costs — real per-second numbers

Prospolabs is pay-per-generation in USD. No seats, no subscription, no token math — you top up from $5 and spend it per clip. Pricing is per second of finished video, and native audio is included at no extra charge. The retail figures below are the standard rate; the launch price is roughly 40% off.

• Seedance 2.0 — 480p $0.09/sec (retail $0.15), 720p $0.18/sec (retail $0.30), 1080p $0.41/sec (retail $0.683).
• Seedance 2.0 Fast — 480p $0.07/sec (retail $0.117), 720p $0.15/sec (retail $0.25).

Put that against a real launch. A 5-second 720p hook on Seedance 2.0 Fast is 5 × $0.15 = $0.75. A 5-second 1080p hero cut on Seedance 2.0 is 5 × $0.41 = $2.05. Draft 30 hook variants at $0.75 and you've spent $22.50 — less than the rush fee on a single human UGC deliverable, with no per-seat tooling cost on top. See the full breakdown on per-generation pricing or stack it against other models on compare model prices.

The batch and variant workflow

The reason to generate ads instead of book them is volume. You don't know which hook converts until it runs, so the smart move is to flood the test with cheap variations and let the data pick the winner.

• Vary the hook, not the whole brief. Keep the creator, product, and room fixed; change only the opening line and tone. That isolates which message lands.
• Draft on Fast, finish on standard. Run every variant at 480p or 720p on Seedance 2.0 Fast, then re-generate only the top performers at 1080p on Seedance 2.0.
• Lock the creator with reference-to-video. Once a face works, cite it as @Image1 across new prompts so the same person carries your whole campaign — different hooks, consistent persona.
• Script the loop via the API. Fire each variant as its own POST /v1/generate, collect the generation_id values, and pull each output_url as it finishes — ideal for generating 20 to 50 clips per launch unattended.
• Re-frame for placement. Generate the source clip, then cut vertical for TikTok and Reels and keep the wider crop for in-feed placements.

When you're ready to extend beyond talking-head ads, the same engine animates product shots into hero clips — see product videos from a photo — or browse the full lineup on explore all video models.

FAQ