I have long been fascinated by stepping inside storybook pictures, a childhood spark that now connects to Google DeepMind’s Genie 4, an AI system that creates real‑time, interactive 3‑D worlds from a single static 2‑D image.
Genie 4 skips all that.
Audio and visuals born together.
Controls learned from observation.
A reasoning loop (powered by Gemini).
Training gym for physical robots.
Genie 1: 2D retro platformers at 1 frame per second. Silent and clunky.
Genie 4 offers Native sound, spatial audio, and live world-editing.