MilliVid, a long-context video generation method without retrieval heuristics, accepted at NeurIPS

CSProfKGD · x · 2026-09-26

The MilliVid team announced their new long-context video generation method has been accepted at NeurIPS. MilliVid produces videos that stay consistent over long time spans without relying on retrieval heuristics or 3D maps, targeting the notoriously hard problem of temporal coherence in extended generation.

Original post →

More from Multimodal

Multimodal channel →