Turn stunning real-world or AI-generated images into Minecraft builds — powered by Hunyuan 2.1 and voxel conversion pipelines.
Click to watch the Eiffel Tower built in Minecraft using an image and Hunyuan's imagination.
Minecraft agents are getting better at chopping trees and mining but when it comes to building realistic, beautiful, and creative structures, they fail.
Inspired by projects like Claude building the Eiffel Tower (poorly 😬), this project bridges the gap between vision models and blocky reality.
-
Image Input
Provide a real-world or AI-generated image of a structure. -
Hunyuan 2.1 Vision Model
We extract structural and spatial data using Hunyuan. -
Voxelization
Convert image → 3D voxel matrix (supports STL/OBJ pipelines or direct voxel inference). -
Minecraft Block Mapping
Map voxel materials to Minecraft blocks intelligently.
--
[ ] Implement cluster detection unsupervised algorithm to automatically detect colors for generated textures
[ ] Develop an algorithm to convert the color hues to minecraft blocks like orange = pumpkin block