ComfyUI Video Generation Hurdle: Forced Speech Censorship Ruins Output

YouNoTypey · reddit · 2026-08-06

A user reported an issue with the default Ref2Vid template in ComfyUI, noting that the required MiniMax model enforces strict speech censorship. This filter skips over flagged dialogue segments without warning, ruining the humor and flow of videos that take significant time to render. The user is currently looking for alternative models without these restrictions.

Original post →

More from Multimodal

Multimodal channel →