What each reference type does
- Images: identity and style — a character, a product, a costume, a colour palette.
- Videos: motion and pacing — how something moves or how a camera travels.
- Audio: rhythm and ambience — a beat to cut to or a room sound to match.
Why reference runs stop at 15 seconds
Wan bills the normalized length of your reference videos alongside the output and requires the two together to stay within thirty seconds. That is why text and first-frame runs go to 30 seconds while reference runs stop at 15. Billed duration rounds up to the next second.
Keeping a character consistent
Use two to four clear images of the same character from different angles, then describe the new action. Avoid mixing references of different characters unless the shot needs both.
The woman from the reference images walks through a neon-lit market at night, handheld camera following at shoulder height, rain on the stalls, distant chatter and sizzling food.