Motion from a reference
A series needs continuity: the same character in the same jacket, the same product on the same table, the same grade across five clips. Reference images do that. Pick up to nine stills from your creations — a face, a product shot, a style frame — and Text to Video keeps them consistent while the motion changes.
References and a first frame cannot be combined, so use references when the scene is new and Photo to Video when the exact frame matters. A strong reference set is three images: the subject, the product and one frame that sets the look.
Starter prompt
“The same character walks out of a subway station into evening rain, coat collar up, neon reflections, steady tracking shot”
What this solves
- A character who looks the same in every episode
- A product that keeps its label across five ads
- A visual style carried from a still into motion
- Series content that cuts together as one show
The steps
Make the reference stills first
A character sheet, a product shot and a style frame are the ideal set.
Original Character MakerOpen Text to Video and pick them as references
Up to nine from your creations; the engine reads them alongside the prompt.
Text to VideoChange only the action between clips
Keep the references and the setting words fixed for continuity.
Text to VideoQuestions
How many references should I use?
Three to five. One subject, one product and one style frame cover most series; more references dilute each one.
Why did my references get dropped?
A first frame was set at the same time. The engine takes one or the other — clear the first frame to use references.
Do references work for a real person?
For a consistent real face, build an AI influencer instead; it attaches face crops automatically and handles consent.
Motion from a reference
Open with these settings