Paper summary
Paper summary: Movie Gen: A Cast of Media Foundation Models
Meta's Movie Gen is a set of foundation models for 1080p HD video with synchronised audio, plus personalisation and instruction-based editing; the largest video model is a 30B Transformer.
- The largest video model is a 30B-parameter Transformer with a 73K video-token context, enough for 16 seconds at 16 fps.
- The report covers the whole stack: architecture, latents, objectives, data curation, evaluation, parallelism and inference.