Flow matching transformer 1

Next-generation AI cinematic video generator with natively integrated, synchronized audio. Muse Video AI Video Generator, developed by Meta Superintelligence Labs (MSL), is a next-generation generative video model designed to turn a line of text, a still image, or a rough video clip into a polished, cinema-grade shot. Unlike traditional models that require separate audio production, Muse Video features native synchronized audio generation, creating sound effects, ambience, and lip-synced dialogue in the same pass as the visuals. Built on a Latent Transformer architecture with Flow Matching and Spatial-Temporal Attention, it provides advanced capabilities like grounded physics, shot-to-shot character consistency, clip remixing, and directable camera movements to meet the needs of filmmakers, marketers, and design professionals.