A generative workflow that transforms massing models into immersive architectural visions. Using GANs and NeRFs, Air-Chitectural Dreams blends video diffusion, machine vision, and 3D generation to reimagine form and materiality.

Captured from all angles, massing models are processed into neural radiance fields, then converted into 3D meshes with AI-mapped textures. Advanced video-to-video diffusion refines details while preserving overall structure, creating fluid, dreamlike interpretations.

By dissolving rigid boundaries, the project explores new modes of architectural visualization. GANs introduce organic unpredictability, while NeRF technology enables interactive spatial experiences, reimagining architectural visualization.