Optional: upload a character reference, a product reference, or both. Use this when you want the same person, the same product, or both to stay consistent across scenes.
Best default: leave this blank and let the app generate a full story arc. Only use overrides if you want to force specific spoken beats.
Controls visible subtitle/text-overlay preference inside the generated video.
Best-practice guide for cleaner videos
Recommended production setups • 30 sec ad = 3 scenes × 10 sec • 40 sec ad = 4 scenes × 10 sec • 10 sec total is best for testing, not final production
Best practice for spoken lines • Start with a real story, not disconnected slogans • Best default: leave scene lines blank so the app writes a connected narrative • If overriding, make each line continue the previous one naturally • For 5 sec scenes: keep each line tight and conversational • For 10 sec scenes: use 1 medium sentence or 2 short natural phrases
Good scene structure • 3 scenes: Hook → Benefit/Proof → CTA • 4 scenes: Hook → Problem → Solution/Proof → CTA
Avoid • long run-on lines • too many claims in one clip • difficult brand phrasing / jargon • trying to force a full ad script into one scene
Continuity mode • Use Character reference image when you want the same person across scenes • Continuity mode may use a stronger internal routing path than a normal single-clip render
Project type
For long ads: Total final duration = number of scenes × per-scene duration. Recommended production defaults: 3 scenes × 10 sec = 30 sec, or 4 scenes × 10 sec = 40 sec. Quick test option: 2 scenes × 5 sec = 10 sec.
Render controls
Loading model options…
No job yet.
Output
Generate a brief to see the script, concepts, render config, prompt, and voiceover draft.