Time2025
Program
HDR.INC.
Leader:
Matt Goldsberry
Exhibition:
KANEKO Museum, Omaha
Blocktown Builder is an experimental 3D building sandbox inspired by the playful city-making spirit of Townscaper. It invites users to construct architectural scenes on a procedurally generated hexagonal world, combining simple controls with richly expressive outcomes. By placing and removing blocks, switching between cube and slope geometries, and adjusting spatial compositions in real time, players can quickly iterate and watch stylized architectural forms emerge.
At its core, the system integrates custom procedural hex-grid generation, voxel-based building logic, and dynamic mesh construction, with Three.js powering rendering and interaction. Beyond basic building tools, users can manipulate color, brightness, transparency, and dynamic sunlight parameters, transforming a scene from a warm daylight study to a dramatic low-light composition. The result is less a rigid simulator and more a playful design instrument—part game, part architectural sketchpad.
A defining feature of the project is its physical interface integration. Blocktown Builder connects to an Arduino-based control panel via the browser’s Web Serial API, allowing knobs and buttons to directly manipulate in-game parameters. Users navigate and build within the virtual space using a gaming controller, reinforcing the embodied, spatial quality of interaction.
The rendering controller itself is housed inside a restored 1984 TI-99/4A computer, bringing a vintage computing artifact back to life. An Arduino reads the original 24-pin keyboard matrix and bridges it to the modern system, while additional knobs and buttons control rendering parameters in real time. By embedding AI-driven visualization within retro hardware, the project reframes an obsolete object as a contemporary creative interface.
Blocktown Builder further extends into AI-assisted visualization. A local Node.js/Express proxy server captures scene context and communicates with FLUX Kontext Pro to generate rendered architectural images from prompts and style presets. This transforms the system into both an interactive sandbox and a concept-rendering workflow—bridging game-like exploration, procedural design, physical computing, and generative AI within a single creative pipeline.
The D3 (Data-Driven Design) group, where I work, is HDR’s digital innovation hub. It focuses on computational design, data visualization, and the development of AI-powered design tools. The team brings together architects, programmers, and data scientists to create cutting-edge products such as AI CAD plugins and design decision-support platforms, making the design process more efficient, intelligent, and sustainable.