flux3 appears to keep voices and accents consistent across a 20-second video

cocktailpeanut · x · 2026-07-26

A user says flux3 can speak Japanese, American-accent English, and Japanese-accent English within the same 20-second video, which they call the first time they have seen that kind of voice and language consistency.

The post quotes another user saying FLUX 3 is very good at maintaining characters and voices throughout a 20-second video, with the prompt linked below.

Original post →

More from Multimodal

Multimodal channel →