LayerT2V: A Unified Multi-Layer Video Generation Framework

LayerT2V introduces a breakthrough multi-layer video generation framework that produces editable layered video components (background, foreground layers with alpha mattes) in a single inference pass. The system addresses professional workflow limitations of current text-to-video models by enabling semantic consistency across layers and introduces VidLayer, the first large-scale dataset for multi-layer video generation.