Machine learning approach for transferring architectural knowledge to design variants

Generative design has since its creation allowed many architects from around the world to get visions of desired buildings. Our project takes generative design to the next level by introducing the power of AI. Based on Conditional GAN we generate realistic floorplans conditioned by constraints of shape and energy consumption. The information of the generated floorplans is extracted to allow a conversion in 3D models. The architect specifies his constraints and a variety of fitting floorplans are created than he can visualize as a base of further work.     «

Generative design has since its creation allowed many architects from around the world to get visions of desired buildings. Our project takes generative design to the next level by introducing the power of AI. Based on Conditional GAN we generate realistic floorplans conditioned by constraints of shape and energy consumption. The information of the generated floorplans is extracted to allow a conversion in 3D models. The architect specifies his constraints and a variety of...     »