World models and visual systems push toward controllable, grounded interaction

BirdDreams couples perspective video with metric-scale BEV for driving-world simulation and trajectory control, while project-page claims await independent validation. WorldDraft probes executable 4D dynamics versus appearance fidelity; Ropedia Xperience 2 contributes large-scale physical-AI scans and recordings, and TTB and Open Vision Banana extend the thread toward faster view synthesis and open generator-based perception. Budget3D is a relevant image-to-3D release lead, but its available catalog material supplies no quality or efficiency evidence.