Why it matters
Video generation is frontier. Understanding shapes capability trend.
Advertisement
The architecture
Latent video: temporal 3D VAE.
DiT with spatial + temporal attention.
Long generation: chained clips.
Advertisement
How it works end to end
3D VAE: compress video to latents.
DiT with 3D attention.
Sampling: 25-50 steps per clip.
Long: consistency across chained clips.