Why UGC works — and why it's hard to fake
UGC (user-generated-content) style ads outperform polished spots on social because they don't look like ads: one person, direct to camera, telling you about something like a friend would. The catch: the style is a discipline. Handheld energy, imperfect lighting, testimonial pacing — get one wrong and it reads as corporate cosplay.
The UGC format preset encodes that discipline. Choose it and the director-brain plans selfie-style MCU/CU framing, splits the testimonial into 1–2 sentence beats, and writes prompts that say "smartphone footage, creator vlog style" instead of "cinematic film still."
Step 1 — Set up the brand (once, reusable forever)
Before creating the show, build a Brand kit in the New Production form (+ New next to the brand-kit select):
- Name: your product or client.
- Brand brief: tone, tagline, CTA, and hard rules. Example: "Playful but premium. Tagline: 'Wake up on the right side.' CTA: 'Link in bio.' Never mention price, never show competitors."
- Image URLs: product shots and logo, one per line.
Two things happen when a show uses the kit: the brief becomes a strict directive to the director-brain (your CTA and tone survive the breakdown), and the images arrive as pre-approved prop canon — your actual product, usable as a reference in any shot.
Step 2 — Create the show
Press Start this project below. Check the settings: format UGC, aspect 9:16, and attach your brand kit. Roll camera on the script.
In the breakdown, verify the shape: no establishing wides, one presenter, dialogue chunked into short beats, your product as canon. That's the format preset working.
Step 3 — Canon: the presenter is your ad
Spend your care on the presenter's canon image. UGC lives or dies on the person feeling real: specific, casual, imperfect. "Mid-20s, curly hair in a claw clip, hoodie, ring-light glow" — not "beautiful influencer."
Point the product shots at your brand-kit prop via the Refs chips, so the real bottle/device/app appears — not a hallucinated cousin of it.
Step 4 — Generate, then give it a voice
After the shots render, hit 🎙 Narrate all. Every dialogue beat gets a voiceover clip in the show's cast voice (pick one in the Export phase's Audio panel — try Aria or Liam for creator energy). Preview each line on its shot card; Re-narrate any read you don't like. Editing a line's text automatically invalidates its old audio.
Lip-sync note: the video models animate the presenter talking, and the TTS track carries the words. They're close but not phoneme-perfect — which, for UGC, usually reads as authentic. For hero lines where sync matters, keep the on-screen beat short.
Step 5 — Captions and the CTA
Social autoplay is muted. Captions aren't optional.
- On each shot card, set Caption / CTA overlay — default to the dialogue line, punchier if you can.
- On the final shot, the CTA: "Link in bio →".
- Export gives you a standard .srt file, and the overlays preview live on the timeline monitor.
Step 6 — Assemble and ship
In Export: generate a music bed if the ad wants one (keep it low-key: "soft lo-fi beat, unobtrusive"), then Assemble preview cut — narration and music are mixed into the final render, already vertical because you shot 9:16.
The UGC checklist
- Hook lands in the first shot (a claim or a "wait, what?" visual)
- Presenter feels like a person, not a render
- Real product (brand-kit ref), not a lookalike
- Every beat captioned; CTA on the last frame
- Under 60 seconds; trim the flabby tails off clips on the timeline
Next lesson: Cut a trailer — spectacle models and accelerating pacing.