Drag to rotate the pipeline · click a stage button to focus it
What each stage actually does
1. Script & Hook (LLM)
A language model like Claude drafts the hook, angle, and 30–60 second script. Good prompts include the product, target persona, one pain point, and a pattern-interrupt opening line. Cost: fractions of a cent per script on API pricing, or a ~$20/mo subscription.
2. Character Lock (Image Model)
An image model generates one reference frame that locks the character’s face, outfit, and setting. Consistency here is what makes the final video feel like one real creator instead of a shapeshifting avatar.
3. Video Generation
The reference frame plus script beats go to a video model. This is the priciest stage — typically $0.10–$0.50 per generated second, so a 30s ad can cost $3–$15 in raw generation before retries.
4. Voice (TTS)
Text-to-speech clones or synthesizes a natural voice. Roughly $0.05–$0.30 per minute on modern TTS APIs. Pacing and breath sounds matter more than raw voice quality for ad performance.
5. Edit & Captions
Auto-editors add captions, jump cuts, b-roll, and platform-safe aspect ratios. Bundled tools run $15–$30/mo. Captions alone lift completion rates because most feed video is watched muted.
Reality check
- Retries are the hidden cost: budget 2–3 generations per usable video.
- Ad platforms require AI-content disclosure in many regions.
- Human UGC still wins on authenticity for high-trust products.
- The $5 figure assumes volume — low volume means subscriptions dominate.
Worked example
Suppose you ship 30 videos/month on a $140 tool stack (LLM $20 + image $20 + video credits $70 + TTS $10 + editor $20). That is $4.67 per video. A traditional UGC creator averages $60–$150 per video; at $85 you would pay $2,550 for the same 30 videos — an 18× difference. But if you only make 5 videos, the same stack costs $28 per video: volume is what makes the math work.
| Stage | Typical tool cost | Share of budget |
|---|---|---|
| Script (LLM) | $20/mo | ~14% |
| Character image | $20/mo | ~14% |
| Video generation | $70/mo credits | ~50% |
| Voice / TTS | $10/mo | ~7% |
| Editing & captions | $20/mo | ~14% |