Reference Forcing, per its authors, cuts drift in long AI video
Reference Forcing reuses a clip's first frames to limit drift in long video; the authors report top consistency scores on MovieGenBench.
Text-to-video and image-to-video models, AI video editing, avatars and lip-sync, open video models, prices and benchmarks.
Reference Forcing reuses a clip's first frames to limit drift in long video; the authors report top consistency scores on MovieGenBench.
Meituan's LongCat-Video-Avatar 1.6 places second in a human-preference Arena, behind OmniHuman-2.0, and targets videos over 10 minutes.
A new reward model for instruction-guided video editing stops picking motion-suppressed edits on VE-Bench, according to its authors' own results.
A new benchmark of 222 experiment pairs finds ten text-and-image-to-video models rarely get both outcomes of a science experiment right.
An arXiv paper presents MemCrafter, an agentic prompt-refinement method for memorable text-to-video, and reports gains over Seedance-2.5 and other baselines.
A paper's visual harness for coding agents scores 58.6% on a 100-task video editing benchmark; we review ablations, MIT licence and API costs.