ALEEEHU/World-Simulator

Organizes multimodal generative models across dimensional progression—2D images, videos, 3D geometry, and 4D spatiotemporal content—within a unified theoretical framework rather than treating modalities independently. Complements the survey with curated Text-to-X resources covering SOTA methods, accepted papers, and implementation codes across specialized domains like human motion synthesis and 3D scene generation. Maintains active community contributions through structured PR workflows and includes technical glossaries defining core concepts like NeRF, 3DGS, and diffusion-based generation paradigms.

332 stars.