Generating Controllable Minecraft Worlds by Training Diffusion Models on Billions of Cubes
tkasasagi · x · 2026-07-30
Sakana AI, in collaboration with NYU, released Dream-Cubed, a new research project. The team harvested chunks of procedural and human-authored Minecraft worlds to train diffusion models.
The model enables highly controllable world generation:
- Autocompletion: It can plausibly autocomplete partially-built environments based on rough sketches (e.g., connecting a snowy biome to a mountain and an ocean).
- Flexible shaping: Allows users to bend naturalistic worlds into compelling and uncanny shapes.
The research shows that similar to tokens in language modeling, "cubes" in video games serve as a natural compositional unit for scalable training with large transformers.
More from Research
- Transforming quadruped robot runs on real hardware: per-pose RL policies in Isaac switched on morph — _Stocko_ · 2026-10-16
- KinoGen synthesizes robot loco-manipulation reference motions directly via kinematically grounded video generation — ChongZzZhang · 2026-10-16
- Science Robotics paper introduces F³, predictive event-based vision beyond optical flow — CSProfKGD · 2026-10-16
- VideoGen-Agent: An 8B Multimodal Agent Turns Video Generation Into a Tool-Use Loop — jiqizhixin · 2026-10-16
- Comprehension Audits: New Paper Proposes a Gate for AI-Written Code at Frontier Labs — ronbodkin · 2026-10-16
- Not Boring #215: Cammarata's Clean Networks Claim, Tesla's Flying Roadster, Tau Cut 73% — Not Boring (Packy McCormick) · 2026-10-16