We propose Prism, a dynamic sparse attention framework for natively training joint video-audio generation models at 2K.
- diffusion-models
- high-resolution
- sparse-attention
- video-generation
Repository Index
47,133 repositories indexed
1–2 of 2 · Most stars
[arXiv 2026] Unified Audio-Video Creation via Progressive Training and Asymmetric Scheduling
- audio-to-video
- joint-audio-video-generation
- unified-model
- video-to-audio