Sakana AI Unveils Dream-Cubed: Minecraft Generative Model Trained on Billions of Voxels
SakanaAILabs · x · 2026-07-30
Sakana AI, in collaboration with NYU, released Dream-Cubed, exploring controllable generative modeling in Minecraft.
- Massive Dataset: Constructed a large-scale dataset of Minecraft worlds at voxel resolution, comprising tens of billions of tokens from procedural terrain and human-authored maps.
- 3D Diffusion Study: Conducted the first large-scale study of 3D diffusion models for voxel generation, analyzing discrete vs. continuous formulations, data compositions, and architectures.
- Interactive Workflows: Operating directly in the space of blocks, the model supports interactive user workflows like inpainting and outpainting from user-authored blocks.
- Evaluation & Open Source: Adapted the FID metric for semantic evaluation and validated quality via human preference studies. The dataset, code, and pretrained models are fully open-sourced.
More from Multimodal
- Seedance 2.5 workflow: character reference sheets + structured prompt yields a 30s single-take fake tourist video — techhalla · 2026-10-16
- Magnific One prompt walkthrough: photoreal misty castle viewpoint rendered at 9:16 1K — techhalla · 2026-10-16
- Smoky Velvet Noir: a reusable image-gen prompt template with three moody examples — LudovicCreator · 2026-10-16
- VideoGen-Agent: An 8B Multimodal Agent Turns Video Generation Into a Tool-Use Loop — jiqizhixin · 2026-10-16
- Redditor builds impressive seaplane sim project with Sonnet 5.5 + Blender — Wise_Shame_2853 · 2026-10-16
- Redditor automates LoRA dataset captioning for 300 photos with Muse AI — Wide_Director_8897 · 2026-10-16