Diffusion Tectonics explores the application of generative AI in shaping proto-architectural tectonic fragments.

A central focus is the training of AI diffusion models, a process that involves incorporating existing design knowledge from the lab and integrating new design data from experimental machine learning methods. This comprehensive dataset covers tectonic logic, structural systems, fabrication techniques, and aesthetic considerations.

Diffusion Tectonics explores the the capacity of machine learning to negotiate between design intention described throuh generative algorithms, digital modeling, drawing, and text prompts to explore diverse design expressions.

The research encompasses several key aspects:

• Data Generation: Crafting data that encapsulates design concepts into images, forming the basis for AI learning.

• Model Training: The core of Diffusion Tectonics, where AI diffusion models are honed, blending human creativity and computational capabilities.

• Image to 3D Conversion: Bridging the gap between images and 3D models to enable a more cohesive design process.

• Hybrid Model Fusion: Exploring hybrid models that merge AI-generated ideas with conventional design approaches.

• Techniques: Employing various techniques including LoRA models, Diffusion Models, NeRF, and Stable Diffusion to advance architectural possibilities.