Atlas · GenAI 2026
Video Generation
Text/image-to-video diffusion and transformer models (Sora, Veo, Runway) with spatiotemporal consistency.
conceptPeak: 2025Generative ArchitecturesAI consensus: 0/3
Prerequisites
- hardDiffusion Models
Most video generators are spatiotemporal diffusion models.
- mediumMultimodal AI
Conditioning spans text, image and time.
Recommended reference
Video generation models as world simulators — OpenAI (Sora technical report), 2024